diff --git a/modin/conftest.py b/modin/conftest.py
index d860807ec19..417f6fe70c3 100644
--- a/modin/conftest.py
+++ b/modin/conftest.py
@@ -306,43 +306,31 @@ def pytest_runtest_call(item):
 
 
 @pytest.fixture(scope="class")
-def TestReadCSVFixture():
+def TestReadCSVFixture(tmp_path_factory):
+    tmp_path = tmp_path_factory.mktemp("TestReadCSVFixture")
     filenames = []
-    files_ids = [
-        "test_read_csv_regular",
-        "test_read_csv_blank_lines",
-        "test_read_csv_yes_no",
-        "test_read_csv_nans",
-        "test_read_csv_bad_lines",
-    ]
     # each xdist worker spawned in separate process with separate namespace and dataset
-    pytest.csvs_names = {file_id: get_unique_filename() for file_id in files_ids}
+    pytest.csvs_names = {}
     # test_read_csv_col_handling, test_read_csv_parsing
-    _make_csv_file(filenames)(
-        filename=pytest.csvs_names["test_read_csv_regular"],
-    )
+    pytest.csvs_names["test_read_csv_regular"] = _make_csv_file(data_dir=tmp_path)()
     # test_read_csv_parsing
-    _make_csv_file(filenames)(
-        filename=pytest.csvs_names["test_read_csv_yes_no"],
+    pytest.csvs_names["test_read_csv_yes_no"] = _make_csv_file(data_dir=tmp_path)(
         additional_col_values=["Yes", "true", "No", "false"],
     )
     # test_read_csv_col_handling
-    _make_csv_file(filenames)(
-        filename=pytest.csvs_names["test_read_csv_blank_lines"],
+    pytest.csvs_names["test_read_csv_blank_lines"] = _make_csv_file(data_dir=tmp_path)(
         add_blank_lines=True,
     )
     # test_read_csv_nans_handling
-    _make_csv_file(filenames)(
-        filename=pytest.csvs_names["test_read_csv_nans"],
+    pytest.csvs_names["test_read_csv_nans"] = _make_csv_file(data_dir=tmp_path)(
         add_blank_lines=True,
         additional_col_values=["<NA>", "N/A", "NA", "NULL", "custom_nan", "73"],
     )
     # test_read_csv_error_handling
-    _make_csv_file(filenames)(
-        filename=pytest.csvs_names["test_read_csv_bad_lines"],
+    pytest.csvs_names["test_read_csv_bad_lines"] = _make_csv_file(data_dir=tmp_path)(
         add_bad_lines=True,
     )
-
+    filenames.extend(pytest.csvs_names.values())
     yield
     # Delete csv files that were created
     teardown_test_files(filenames)
@@ -350,19 +338,14 @@ def TestReadCSVFixture():
 
 @pytest.fixture
 @doc(_doc_pytest_fixture, file_type="csv")
-def make_csv_file():
-    filenames = []
-
-    yield _make_csv_file(filenames)
-
-    # Delete csv files that were created
-    teardown_test_files(filenames)
+def make_csv_file(tmp_path):
+    yield _make_csv_file(data_dir=tmp_path)
 
 
 def create_fixture(file_type):
     @doc(_doc_pytest_fixture, file_type=file_type)
-    def fixture():
-        func, filenames = make_default_file(file_type=file_type)
+    def fixture(tmp_path):
+        func, filenames = make_default_file(file_type=file_type, data_dir=tmp_path)
         yield func
         teardown_test_files(filenames)
 
@@ -476,20 +459,18 @@ def _sql_connection(filename, table=""):
 
 
 @pytest.fixture(scope="class")
-def TestReadGlobCSVFixture():
-    filenames = []
+def TestReadGlobCSVFixture(tmp_path_factory):
+    tmp_path = tmp_path_factory.mktemp("TestReadGlobCSVFixture")
 
     base_name = get_unique_filename(extension="")
-    pytest.glob_path = "{}_*.csv".format(base_name)
-    pytest.files = ["{}_{}.csv".format(base_name, i) for i in range(11)]
+    pytest.glob_path = str(tmp_path / "{}_*.csv".format(base_name))
+    pytest.files = [str(tmp_path / "{}_{}.csv".format(base_name, i)) for i in range(11)]
     for fname in pytest.files:
         # Glob does not guarantee ordering so we have to remove the randomness in the generated csvs.
-        _make_csv_file(filenames)(fname, row_size=11, remove_randomness=True)
+        _make_csv_file(data_dir=tmp_path)(fname, row_size=11, remove_randomness=True)
 
     yield
 
-    teardown_test_files(filenames)
-
 
 @pytest.fixture
 def get_generated_doc_urls():
diff --git a/modin/experimental/pandas/test/test_io_exp.py b/modin/experimental/pandas/test/test_io_exp.py
index 4bfd8c9b7ab..dd7da32bd12 100644
--- a/modin/experimental/pandas/test/test_io_exp.py
+++ b/modin/experimental/pandas/test/test_io_exp.py
@@ -41,7 +41,7 @@
 def test_from_sql_distributed(tmp_path, make_sql_connection):
     filename = "test_from_sql_distributed.db"
     table = "test_from_sql_distributed"
-    conn = make_sql_connection(tmp_path / filename, table)
+    conn = make_sql_connection(str(tmp_path / filename), table)
     query = "select * from {0}".format(table)
 
     pandas_df = pandas.read_sql(query, conn)
@@ -73,7 +73,7 @@ def test_from_sql_distributed(tmp_path, make_sql_connection):
 def test_from_sql_defaults(tmp_path, make_sql_connection):
     filename = "test_from_sql_distributed.db"
     table = "test_from_sql_distributed"
-    conn = make_sql_connection(tmp_path / filename, table)
+    conn = make_sql_connection(str(tmp_path / filename), table)
     query = "select * from {0}".format(table)
 
     pandas_df = pandas.read_sql(query, conn)
@@ -134,7 +134,7 @@ def test_read_csv_without_glob(self):
                     storage_options={"anon": True},
                 )
 
-    def test_read_csv_glob_4373(self):
+    def test_read_csv_glob_4373(self, tmp_path):
         columns, filename = ["col0"], "1x1.csv"
         df = pd.DataFrame([[1]], columns=columns)
         with (
@@ -142,7 +142,7 @@ def test_read_csv_glob_4373(self):
             if Engine.get() == "Dask"
             else contextlib.nullcontext()
         ):
-            df.to_csv(filename)
+            df.to_csv(str(tmp_path / filename))
 
         kwargs = {"filepath_or_buffer": filename, "usecols": columns}
         modin_df = pd.read_csv_glob(**kwargs)
@@ -203,9 +203,6 @@ def _pandas_read_csv_glob(path, storage_options):
     )
 
 
-test_default_to_pickle_filename = "test_default_to_pickle.pkl"
-
-
 @pytest.mark.skipif(
     Engine.get() not in ("Ray", "Unidist", "Dask"),
     reason=f"{Engine.get()} does not have experimental API",
@@ -245,9 +242,9 @@ def _pandas_read_csv_glob(path, storage_options):
 )
 @pytest.mark.parametrize("compression", [None, "gzip"])
 @pytest.mark.parametrize(
-    "filename", [test_default_to_pickle_filename, "test_to_pickle*.pkl"]
+    "filename", ["test_default_to_pickle.pkl", "test_to_pickle*.pkl"]
 )
-def test_distributed_pickling(filename, compression):
+def test_distributed_pickling(tmp_path, filename, compression):
     data = test_data["int_data"]
     df = pd.DataFrame(data)
 
@@ -257,11 +254,13 @@ def test_distributed_pickling(filename, compression):
 
     with (
         warns_that_defaulting_to_pandas()
-        if filename_param == test_default_to_pickle_filename
+        if filename_param == "test_default_to_pickle.pkl"
         else contextlib.nullcontext()
     ):
-        df.to_pickle_distributed(filename, compression=compression)
-        pickled_df = pd.read_pickle_distributed(filename, compression=compression)
+        df.to_pickle_distributed(str(tmp_path / filename), compression=compression)
+        pickled_df = pd.read_pickle_distributed(
+            str(tmp_path / filename), compression=compression
+        )
     df_equals(pickled_df, df)
 
     pickle_files = glob.glob(filename)
diff --git a/modin/pandas/test/test_io.py b/modin/pandas/test/test_io.py
index 6f1aa00c13a..fd801a61d12 100644
--- a/modin/pandas/test/test_io.py
+++ b/modin/pandas/test/test_io.py
@@ -52,7 +52,6 @@
 from modin.utils import to_pandas
 
 from .utils import (
-    COMP_TO_EXT,
     check_file_leaks,
     create_test_dfs,
     default_to_pandas_ignore_string,
@@ -268,42 +267,38 @@ class TestCsv:
     def test_read_csv_delimiters(
         self, make_csv_file, sep, delimiter, decimal, thousands
     ):
-        with ensure_clean(".csv") as unique_filename:
-            make_csv_file(
-                filename=unique_filename,
-                delimiter=delimiter,
-                thousands_separator=thousands,
-                decimal_separator=decimal,
-            )
-
-            eval_io(
-                fn_name="read_csv",
-                # read_csv kwargs
-                filepath_or_buffer=unique_filename,
-                delimiter=delimiter,
-                sep=sep,
-                decimal=decimal,
-                thousands=thousands,
-            )
+        unique_filename = make_csv_file(
+            delimiter=delimiter,
+            thousands_separator=thousands,
+            decimal_separator=decimal,
+        )
+        eval_io(
+            fn_name="read_csv",
+            # read_csv kwargs
+            filepath_or_buffer=unique_filename,
+            delimiter=delimiter,
+            sep=sep,
+            decimal=decimal,
+            thousands=thousands,
+        )
 
     @pytest.mark.parametrize(
         "dtype_backend", [lib.no_default, "numpy_nullable", "pyarrow"]
     )
     def test_read_csv_dtype_backend(self, make_csv_file, dtype_backend):
-        with ensure_clean(".csv") as unique_filename:
-            make_csv_file(filename=unique_filename)
+        unique_filename = make_csv_file()
 
-            def comparator(df1, df2):
-                df_equals(df1, df2)
-                df_equals(df1.dtypes, df2.dtypes)
+        def comparator(df1, df2):
+            df_equals(df1, df2)
+            df_equals(df1.dtypes, df2.dtypes)
 
-            eval_io(
-                fn_name="read_csv",
-                # read_csv kwargs
-                filepath_or_buffer=unique_filename,
-                dtype_backend=dtype_backend,
-                comparator=comparator,
-            )
+        eval_io(
+            fn_name="read_csv",
+            # read_csv kwargs
+            filepath_or_buffer=unique_filename,
+            dtype_backend=dtype_backend,
+            comparator=comparator,
+        )
 
     # Column and Index Locations and Names tests
     @pytest.mark.parametrize("header", ["infer", None, 0])
@@ -421,38 +416,34 @@ def test_read_csv_parsing_2(
         names,
         encoding,
     ):
-        with ensure_clean(".csv") as unique_filename:
-            if encoding:
-                make_csv_file(
-                    filename=unique_filename,
-                    encoding=encoding,
-                )
-            kwargs = {
-                "filepath_or_buffer": unique_filename
-                if encoding
-                else pytest.csvs_names["test_read_csv_regular"],
-                "header": header,
-                "skiprows": skiprows,
-                "nrows": nrows,
-                "names": names,
-                "encoding": encoding,
-            }
+        if encoding:
+            unique_filename = make_csv_file(
+                encoding=encoding,
+            )
+        kwargs = {
+            "filepath_or_buffer": unique_filename
+            if encoding
+            else pytest.csvs_names["test_read_csv_regular"],
+            "header": header,
+            "skiprows": skiprows,
+            "nrows": nrows,
+            "names": names,
+            "encoding": encoding,
+        }
 
-            if Engine.get() != "Python":
-                df = pandas.read_csv(**dict(kwargs, nrows=1))
-                # in that case first partition will contain str
-                if df[df.columns[0]][df.index[0]] in ["c1", "col1", "c3", "col3"]:
-                    pytest.xfail(
-                        "read_csv incorrect output with float data - issue #2634"
-                    )
+        if Engine.get() != "Python":
+            df = pandas.read_csv(**dict(kwargs, nrows=1))
+            # in that case first partition will contain str
+            if df[df.columns[0]][df.index[0]] in ["c1", "col1", "c3", "col3"]:
+                pytest.xfail("read_csv incorrect output with float data - issue #2634")
 
-            eval_io(
-                fn_name="read_csv",
-                raising_exceptions=None,
-                check_kwargs_callable=not callable(skiprows),
-                # read_csv kwargs
-                **kwargs,
-            )
+        eval_io(
+            fn_name="read_csv",
+            raising_exceptions=None,
+            check_kwargs_callable=not callable(skiprows),
+            # read_csv kwargs
+            **kwargs,
+        )
 
     @pytest.mark.parametrize("true_values", [["Yes"], ["Yes", "true"], None])
     @pytest.mark.parametrize("false_values", [["No"], ["No", "false"], None])
@@ -640,24 +631,16 @@ def test_read_csv_encoding_976(self):
     @pytest.mark.parametrize("encoding", [None, "latin8", "utf16"])
     @pytest.mark.parametrize("engine", [None, "python", "c"])
     def test_read_csv_compression(self, make_csv_file, compression, encoding, engine):
-        with ensure_clean(".csv") as unique_filename:
-            make_csv_file(
-                filename=unique_filename, encoding=encoding, compression=compression
-            )
-            compressed_file_path = (
-                f"{unique_filename}.{COMP_TO_EXT[compression]}"
-                if compression != "infer"
-                else unique_filename
-            )
+        unique_filename = make_csv_file(encoding=encoding, compression=compression)
 
-            eval_io(
-                fn_name="read_csv",
-                # read_csv kwargs
-                filepath_or_buffer=compressed_file_path,
-                compression=compression,
-                encoding=encoding,
-                engine=engine,
-            )
+        eval_io(
+            fn_name="read_csv",
+            # read_csv kwargs
+            filepath_or_buffer=unique_filename,
+            compression=compression,
+            encoding=encoding,
+            engine=engine,
+        )
 
     @pytest.mark.parametrize(
         "encoding",
@@ -685,15 +668,13 @@ def test_read_csv_compression(self, make_csv_file, compression, encoding, engine
         ],
     )
     def test_read_csv_encoding(self, make_csv_file, encoding):
-        with ensure_clean(".csv") as unique_filename:
-            make_csv_file(filename=unique_filename, encoding=encoding)
-
-            eval_io(
-                fn_name="read_csv",
-                # read_csv kwargs
-                filepath_or_buffer=unique_filename,
-                encoding=encoding,
-            )
+        unique_filename = make_csv_file(encoding=encoding)
+        eval_io(
+            fn_name="read_csv",
+            # read_csv kwargs
+            filepath_or_buffer=unique_filename,
+            encoding=encoding,
+        )
 
     @pytest.mark.parametrize("thousands", [None, ",", "_", " "])
     @pytest.mark.parametrize("decimal", [".", "_"])
@@ -709,56 +690,50 @@ def test_read_csv_file_format(
         escapechar,
         dialect,
     ):
-        with ensure_clean(".csv") as unique_filename:
-            if dialect:
-                test_csv_dialect_params = {
-                    "delimiter": "_",
-                    "doublequote": False,
-                    "escapechar": "\\",
-                    "quotechar": "d",
-                    "quoting": csv.QUOTE_ALL,
-                }
-                csv.register_dialect(dialect, **test_csv_dialect_params)
-                if dialect != "use_dialect_name":
-                    # otherwise try with dialect name instead of `_csv.Dialect` object
-                    dialect = csv.get_dialect(dialect)
-                make_csv_file(filename=unique_filename, **test_csv_dialect_params)
-            else:
-                make_csv_file(
-                    filename=unique_filename,
-                    thousands_separator=thousands,
-                    decimal_separator=decimal,
-                    escapechar=escapechar,
-                    lineterminator=lineterminator,
-                )
-
-            if (
-                (StorageFormat.get() == "Hdk")
-                and (escapechar is not None)
-                and (lineterminator is None)
-                and (thousands is None)
-                and (decimal == ".")
-            ):
-                with open(unique_filename, "r") as f:
-                    if any(
-                        line.find(f',"{escapechar}') != -1 for _, line in enumerate(f)
-                    ):
-                        pytest.xfail(
-                            "Tests with this character sequence fail due to #5649"
-                        )
-
-            eval_io(
-                raising_exceptions=None,
-                fn_name="read_csv",
-                # read_csv kwargs
-                filepath_or_buffer=unique_filename,
-                thousands=thousands,
-                decimal=decimal,
-                lineterminator=lineterminator,
+        if dialect:
+            test_csv_dialect_params = {
+                "delimiter": "_",
+                "doublequote": False,
+                "escapechar": "\\",
+                "quotechar": "d",
+                "quoting": csv.QUOTE_ALL,
+            }
+            csv.register_dialect(dialect, **test_csv_dialect_params)
+            if dialect != "use_dialect_name":
+                # otherwise try with dialect name instead of `_csv.Dialect` object
+                dialect = csv.get_dialect(dialect)
+            unique_filename = make_csv_file(**test_csv_dialect_params)
+        else:
+            unique_filename = make_csv_file(
+                thousands_separator=thousands,
+                decimal_separator=decimal,
                 escapechar=escapechar,
-                dialect=dialect,
+                lineterminator=lineterminator,
             )
 
+        if (
+            (StorageFormat.get() == "Hdk")
+            and (escapechar is not None)
+            and (lineterminator is None)
+            and (thousands is None)
+            and (decimal == ".")
+        ):
+            with open(unique_filename, "r") as f:
+                if any(line.find(f',"{escapechar}') != -1 for _, line in enumerate(f)):
+                    pytest.xfail("Tests with this character sequence fail due to #5649")
+
+        eval_io(
+            raising_exceptions=None,
+            fn_name="read_csv",
+            # read_csv kwargs
+            filepath_or_buffer=unique_filename,
+            thousands=thousands,
+            decimal=decimal,
+            lineterminator=lineterminator,
+            escapechar=escapechar,
+            dialect=dialect,
+        )
+
     @pytest.mark.parametrize(
         "quoting",
         [csv.QUOTE_ALL, csv.QUOTE_MINIMAL, csv.QUOTE_NONNUMERIC, csv.QUOTE_NONE],
@@ -780,26 +755,24 @@ def test_read_csv_quoting(
             not doublequote and quotechar != '"' and quoting != csv.QUOTE_NONE
         )
         escapechar = "\\" if use_escapechar else None
-        with ensure_clean(".csv") as unique_filename:
-            make_csv_file(
-                filename=unique_filename,
-                quoting=quoting,
-                quotechar=quotechar,
-                doublequote=doublequote,
-                escapechar=escapechar,
-                comment_col_char=comment,
-            )
+        unique_filename = make_csv_file(
+            quoting=quoting,
+            quotechar=quotechar,
+            doublequote=doublequote,
+            escapechar=escapechar,
+            comment_col_char=comment,
+        )
 
-            eval_io(
-                fn_name="read_csv",
-                # read_csv kwargs
-                filepath_or_buffer=unique_filename,
-                quoting=quoting,
-                quotechar=quotechar,
-                doublequote=doublequote,
-                escapechar=escapechar,
-                comment=comment,
-            )
+        eval_io(
+            fn_name="read_csv",
+            # read_csv kwargs
+            filepath_or_buffer=unique_filename,
+            quoting=quoting,
+            quotechar=quotechar,
+            doublequote=doublequote,
+            escapechar=escapechar,
+            comment=comment,
+        )
 
     # Error Handling parameters tests
     @pytest.mark.skip(reason="https://github.com/modin-project/modin/issues/6239")
@@ -839,6 +812,7 @@ def test_read_csv_internal(
         low_memory,
         memory_map,
         float_precision,
+        tmp_path,
     ):
         # In this case raised TypeError: cannot use a string pattern on a bytes-like object,
         # so TypeError should be excluded from raising_exceptions list in order to check, that
@@ -866,29 +840,28 @@ def test_read_csv_internal(
             "float_precision": float_precision,
         }
 
-        with ensure_clean(".csv") as unique_filename:
-            if use_str_data:
-                str_delim_whitespaces = (
-                    "col1 col2  col3   col4\n5 6   7  8\n9  10    11 12\n"
-                )
-                eval_io_from_str(
-                    str_delim_whitespaces,
-                    unique_filename,
-                    raising_exceptions=raising_exceptions,
-                    **kwargs,
-                )
-            else:
-                make_csv_file(
-                    filename=unique_filename,
-                    delimiter=delimiter,
-                )
-
-                eval_io(
-                    filepath_or_buffer=unique_filename,
-                    fn_name="read_csv",
-                    raising_exceptions=raising_exceptions,
-                    **kwargs,
-                )
+        if use_str_data:
+            str_delim_whitespaces = (
+                "col1 col2  col3   col4\n5 6   7  8\n9  10    11 12\n"
+            )
+            unique_filename = get_unique_filename(data_dir=tmp_path)
+            eval_io_from_str(
+                str_delim_whitespaces,
+                unique_filename,
+                raising_exceptions=raising_exceptions,
+                **kwargs,
+            )
+        else:
+            unique_filename = make_csv_file(
+                delimiter=delimiter,
+            )
+
+            eval_io(
+                filepath_or_buffer=unique_filename,
+                fn_name="read_csv",
+                raising_exceptions=raising_exceptions,
+                **kwargs,
+            )
 
     # Issue related, specific or corner cases
     @pytest.mark.parametrize("nrows", [2, None])
@@ -1192,21 +1165,13 @@ def wrapped_read_csv(file, method):
     def test_read_csv_file_handle(
         self, read_mode, make_csv_file, buffer_start_pos, set_async_read_mode
     ):
-        with ensure_clean() as unique_filename:
-            make_csv_file(filename=unique_filename)
-
-            with open(unique_filename, mode=read_mode) as buffer:
-                buffer.seek(buffer_start_pos)
-                pandas_df = pandas.read_csv(buffer)
-                buffer.seek(buffer_start_pos)
-                modin_df = pd.read_csv(buffer)
-            if AsyncReadMode.get():
-                # If read operations are asynchronous, then the dataframes
-                # check should be inside `ensure_clean` context
-                # because the file may be deleted before actual reading starts
-                df_equals(modin_df, pandas_df)
-        if not AsyncReadMode.get():
-            df_equals(modin_df, pandas_df)
+        unique_filename = make_csv_file()
+        with open(unique_filename, mode=read_mode) as buffer:
+            buffer.seek(buffer_start_pos)
+            pandas_df = pandas.read_csv(buffer)
+            buffer.seek(buffer_start_pos)
+            modin_df = pd.read_csv(buffer)
+        df_equals(modin_df, pandas_df)
 
     def test_unnamed_index(self):
         def get_internal_df(df):
@@ -1330,21 +1295,19 @@ def _check_relative_io(fn_name, unique_filename, path_arg, storage_default=()):
 # TODO(https://github.com/modin-project/modin/issues/3655): Get rid of this
 # commment once we turn all default to pandas messages into errors.
 def test_read_csv_relative_to_user_home(make_csv_file):
-    with ensure_clean(".csv") as unique_filename:
-        make_csv_file(filename=unique_filename)
-        _check_relative_io("read_csv", unique_filename, "filepath_or_buffer")
+    unique_filename = make_csv_file()
+    _check_relative_io("read_csv", unique_filename, "filepath_or_buffer")
 
 
 @pytest.mark.filterwarnings(default_to_pandas_ignore_string)
 class TestTable:
     def test_read_table(self, make_csv_file):
-        with ensure_clean() as unique_filename:
-            make_csv_file(filename=unique_filename, delimiter="\t")
-            eval_io(
-                fn_name="read_table",
-                # read_table kwargs
-                filepath_or_buffer=unique_filename,
-            )
+        unique_filename = make_csv_file(delimiter="\t")
+        eval_io(
+            fn_name="read_table",
+            # read_table kwargs
+            filepath_or_buffer=unique_filename,
+        )
 
     @pytest.mark.parametrize("set_async_read_mode", [False, True], indirect=True)
     def test_read_table_within_decorator(self, make_csv_file, set_async_read_mode):
@@ -1356,34 +1319,26 @@ def wrapped_read_table(file, method):
             if method == "modin":
                 return pd.read_table(file)
 
-        with ensure_clean() as unique_filename:
-            make_csv_file(filename=unique_filename, delimiter="\t")
+        unique_filename = make_csv_file(delimiter="\t")
 
-            pandas_df = wrapped_read_table(unique_filename, method="pandas")
-            modin_df = wrapped_read_table(unique_filename, method="modin")
+        pandas_df = wrapped_read_table(unique_filename, method="pandas")
+        modin_df = wrapped_read_table(unique_filename, method="modin")
 
         if StorageFormat.get() == "Hdk":
             modin_df, pandas_df = align_datetime_dtypes(modin_df, pandas_df)
 
-            if AsyncReadMode.get():
-                # If read operations are asynchronous, then the dataframes
-                # check should be inside `ensure_clean` context
-                # because the file may be deleted before actual reading starts
-                df_equals(modin_df, pandas_df)
-        if not AsyncReadMode.get():
-            df_equals(modin_df, pandas_df)
+        df_equals(modin_df, pandas_df)
 
     def test_read_table_empty_frame(self, make_csv_file):
-        with ensure_clean() as unique_filename:
-            make_csv_file(filename=unique_filename, delimiter="\t")
+        unique_filename = make_csv_file(delimiter="\t")
 
-            eval_io(
-                fn_name="read_table",
-                # read_table kwargs
-                filepath_or_buffer=unique_filename,
-                usecols=["col1"],
-                index_col="col1",
-            )
+        eval_io(
+            fn_name="read_table",
+            # read_table kwargs
+            filepath_or_buffer=unique_filename,
+            usecols=["col1"],
+            index_col="col1",
+        )
 
 
 @pytest.mark.parametrize("engine", ["pyarrow", "fastparquet"])
diff --git a/modin/pandas/test/utils.py b/modin/pandas/test/utils.py
index ab1251ccf4f..2c73d8caa33 100644
--- a/modin/pandas/test/utils.py
+++ b/modin/pandas/test/utils.py
@@ -1343,9 +1343,9 @@ def generate_dataframe(row_size=NROWS, additional_col_values=None, idx_name=None
     return pandas.DataFrame(data, index=index)
 
 
-def _make_csv_file(filenames):
+def _make_csv_file(data_dir):
     def _csv_file_maker(
-        filename,
+        filename=None,
         row_size=NROWS,
         force=True,
         delimiter=",",
@@ -1365,8 +1365,10 @@ def _csv_file_maker(
         escapechar=None,
         lineterminator=None,
     ):
+        if filename is None:
+            filename = get_unique_filename(data_dir=data_dir)
         if os.path.exists(filename) and not force:
-            pass
+            return None
         else:
             df = generate_dataframe(row_size, additional_col_values)
             if remove_randomness:
@@ -1436,8 +1438,7 @@ def _csv_file_maker(
                     encoding=encoding,
                     **csv_reader_writer_params,
                 )
-            filenames.append(filename)
-            return df
+            return filename
 
     return _csv_file_maker
 
@@ -1493,7 +1494,7 @@ def rotate_decimal_digits_or_symbols(value):
         return tens + ones * 10
 
 
-def make_default_file(file_type: str):
+def make_default_file(file_type: str, data_dir: str):
     """Helper function for pytest fixtures."""
     filenames = []
 
@@ -1523,9 +1524,8 @@ def _create_file(filenames, filename, force, nrows, ncols, func: str, func_kw=No
     }
     extension = file_type_to_extension.get(file_type, file_type)
 
-    def _make_default_file(filename=None, nrows=NROWS, ncols=2, force=True, **kwargs):
-        if filename is None:
-            filename = get_unique_filename(extension=extension)
+    def _make_default_file(nrows=NROWS, ncols=2, force=True, **kwargs):
+        filename = get_unique_filename(extension=extension, data_dir=data_dir)
 
         if file_type == "json":
             lines = kwargs.get("lines")