diff --git a/modules/pubsub/src/main/scala/com/snowplowanalytics/snowplow/enrich/pubsub/Sink.scala b/modules/pubsub/src/main/scala/com/snowplowanalytics/snowplow/enrich/pubsub/Sink.scala
index 39f2dc1c3..c24ec93a7 100644
--- a/modules/pubsub/src/main/scala/com/snowplowanalytics/snowplow/enrich/pubsub/Sink.scala
+++ b/modules/pubsub/src/main/scala/com/snowplowanalytics/snowplow/enrich/pubsub/Sink.scala
@@ -28,8 +28,12 @@ import com.permutive.pubsub.producer.grpc.{GooglePubsubProducer, PubsubProducerC
 import com.snowplowanalytics.snowplow.enrich.common.fs2.{AttributedByteSink, AttributedData, ByteSink}
 import com.snowplowanalytics.snowplow.enrich.common.fs2.config.io.Output
 
+import java.nio.charset.StandardCharsets
+
 object Sink {
 
+  private val maxAttributeBytesLength = 1024
+
   private implicit def unsafeLogger[F[_]: Sync]: Logger[F] =
     Slf4jLogger.getLogger[F]
 
@@ -67,6 +71,13 @@ object Sink {
 
   private def sinkBatch[F[_]: Concurrent: Parallel, A](producer: PubsubProducer[F, A])(records: List[AttributedData[A]]): F[Unit] =
     records.parTraverse_ { r =>
-      producer.produce(r.data, r.attributes)
+      val attributes = dropOversizedAttributes(r)
+      producer.produce(r.data, attributes)
     }.void
+
+  private def dropOversizedAttributes[A](r: AttributedData[A]): Map[String, String] =
+    r.attributes.filter {
+      case (_, value) =>
+        value.getBytes(StandardCharsets.UTF_8).length <= maxAttributeBytesLength
+    }
 }