diff --git a/include/deal.II/base/config.h.in b/include/deal.II/base/config.h.in
index 32b41e4fb026..818a031d06f1 100644
--- a/include/deal.II/base/config.h.in
+++ b/include/deal.II/base/config.h.in
@@ -138,8 +138,6 @@
 #define DEAL_II_COMPILER_VECTORIZATION_LEVEL 0
 #endif
 
-#cmakedefine DEAL_II_HAVE_ARM_NEON
-
 #define DEAL_II_OPENMP_SIMD_PRAGMA @DEAL_II_OPENMP_SIMD_PRAGMA@
 
 
diff --git a/include/deal.II/base/vectorization.h b/include/deal.II/base/vectorization.h
index b43f59a83a02..5a05d48652e2 100644
--- a/include/deal.II/base/vectorization.h
+++ b/include/deal.II/base/vectorization.h
@@ -1003,7 +1003,7 @@ vectorized_transpose_and_store(const bool                            add_into,
 
 #ifndef DOXYGEN
 
-#  if defined(DEAL_II_HAVE_ARM_NEON) && defined(__ARM_NEON)
+#  if DEAL_II_VECTORIZATION_WIDTH_IN_BITS >= 128 && defined(__ARM_NEON)
 
 /**
  * Specialization for double and ARM Neon.
@@ -6313,7 +6313,7 @@ compare_and_apply_mask(const VectorizedArray<double, 2> &left,
 
 #  endif
 
-#  if defined(DEAL_II_HAVE_ARM_NEON) && defined(__ARM_NEON)
+#  if DEAL_II_VECTORIZATION_WIDTH_IN_BITS >= 128 && defined(__ARM_NEON)
 
 template <SIMDComparison predicate>
 DEAL_II_ALWAYS_INLINE inline VectorizedArray<float, 4>