diff --git a/ibis/backends/pyspark/__init__.py b/ibis/backends/pyspark/__init__.py
index 8511f51300e1..c5f921381139 100644
--- a/ibis/backends/pyspark/__init__.py
+++ b/ibis/backends/pyspark/__init__.py
@@ -456,6 +456,8 @@ def _register_in_memory_table(self, op: ops.InMemoryTable) -> None:
 
     def _finalize_memtable(self, name: str) -> None:
         """No-op, otherwise a deadlock can occur when using Spark Connect."""
+        if isinstance(session := self._session, pyspark.sql.SparkSession):
+            session.catalog.dropTempView(name)
 
     @contextlib.contextmanager
     def _safe_raw_sql(self, query: str) -> Any:
diff --git a/ibis/backends/tests/test_client.py b/ibis/backends/tests/test_client.py
index f701f7db3f03..7d1ae751e58b 100644
--- a/ibis/backends/tests/test_client.py
+++ b/ibis/backends/tests/test_client.py
@@ -35,6 +35,7 @@
     PyODBCProgrammingError,
     SnowflakeProgrammingError,
 )
+from ibis.conftest import IS_SPARK_REMOTE
 from ibis.util import gen_name
 
 if TYPE_CHECKING:
@@ -1688,8 +1689,9 @@ def test_insert_into_table_missing_columns(con, temp_table):
 )
 @pytest.mark.notyet(
     ["pyspark"],
+    condition=IS_SPARK_REMOTE,
     raises=AssertionError,
-    reason="likely, but not guaranteed deadlock when using spark connect",
+    reason="likely deadlock when using spark connect",
 )
 def test_memtable_cleanup(con):
     name = ibis.util.gen_name("temp_memtable")