diff --git a/dlt/common/schema/schema.py b/dlt/common/schema/schema.py
index 38f5efc0bc..04ca7f88a1 100644
--- a/dlt/common/schema/schema.py
+++ b/dlt/common/schema/schema.py
@@ -297,7 +297,7 @@ def apply_schema_contract(
         existing_table: TTableSchema = self._schema_tables.get(table_name, None)
 
         # table is new when not yet exist or
-        is_dlt_table = table_name.startswith("_dlt")
+        is_dlt_table = table_name.startswith(self._dlt_tables_prefix)
         should_raise = raise_on_freeze and not is_dlt_table
         is_new_table = not existing_table or self.is_new_table(table_name)
         # check case where we have a new table
diff --git a/tests/pipeline/test_pipeline.py b/tests/pipeline/test_pipeline.py
index aa936a7690..7a70fded1e 100644
--- a/tests/pipeline/test_pipeline.py
+++ b/tests/pipeline/test_pipeline.py
@@ -2280,22 +2280,19 @@ def test_pipeline_with_frozen_schema_contract() -> None:
         destination="duckdb",
     )
 
+    # Create a database schema with table
+    with pipeline.sql_client() as c:
+        dataset = c.fully_qualified_dataset_name()
+        table = f"{c.fully_qualified_dataset_name()}.test_items"
+        conn = c.open_connection()
+        conn.sql(f"CREATE SCHEMA {dataset}")
+        conn.sql(f"CREATE TABLE {table} (id INTEGER PRIMARY KEY, name VARCHAR)")
+
     data = [
         {"id": 101, "name": "sub item 101"},
         {"id": 101, "name": "sub item 102"},
     ]
 
-    pipeline.run(
-        data,
-        table_name="test_items",
-    )
-
-    with pipeline.sql_client() as c:
-        c.execute_sql("DROP TABLE _dlt_loads")
-        c.execute_sql("DROP TABLE _dlt_version")
-        c.execute_sql("DROP TABLE _dlt_pipeline_state")
-        c.execute_sql("TRUNCATE TABLE test_items")
-
     pipeline.run(
         data,
         table_name="test_items",