diff --git a/google/cloud/bigquery/job/query.py b/google/cloud/bigquery/job/query.py
index 23eecfdf79..67955dbb17 100644
--- a/google/cloud/bigquery/job/query.py
+++ b/google/cloud/bigquery/job/query.py
@@ -1342,7 +1342,7 @@ def result(
     def to_arrow(
         self,
         progress_bar_type: str = None,
-        bqstorage_client: "bigquery_storage.BigQueryReadClient" = None,
+        bqstorage_client: Optional["bigquery_storage.BigQueryReadClient"] = None,
         create_bqstorage_client: bool = True,
         max_results: Optional[int] = None,
     ) -> "pyarrow.Table":
@@ -1412,7 +1412,7 @@ def to_arrow(
     # that should only exist here in the QueryJob method.
     def to_dataframe(
         self,
-        bqstorage_client: "bigquery_storage.BigQueryReadClient" = None,
+        bqstorage_client: Optional["bigquery_storage.BigQueryReadClient"] = None,
         dtypes: Dict[str, Any] = None,
         progress_bar_type: str = None,
         create_bqstorage_client: bool = True,
diff --git a/google/cloud/bigquery/table.py b/google/cloud/bigquery/table.py
index c88488b7ba..fee98ce5e5 100644
--- a/google/cloud/bigquery/table.py
+++ b/google/cloud/bigquery/table.py
@@ -36,7 +36,6 @@
 from google.api_core.page_iterator import HTTPIterator
 
 import google.cloud._helpers
-from google.cloud import bigquery_storage
 from google.cloud.bigquery import _helpers
 from google.cloud.bigquery import _pandas_helpers
 from google.cloud.bigquery.exceptions import LegacyBigQueryStorageError
@@ -48,6 +47,8 @@
 from google.cloud.bigquery.encryption_configuration import EncryptionConfiguration
 
 if typing.TYPE_CHECKING:  # pragma: NO COVER
+    from google.cloud import bigquery_storage
+
     # Unconditionally import optional dependencies again to tell pytype that
     # they are not None, avoiding false "no attribute" errors.
     import pandas
@@ -1628,7 +1629,7 @@ def _to_arrow_iterable(self, bqstorage_client=None):
     def to_arrow(
         self,
         progress_bar_type: str = None,
-        bqstorage_client: Optional[bigquery_storage.BigQueryReadClient] = None,
+        bqstorage_client: Optional["bigquery_storage.BigQueryReadClient"] = None,
         create_bqstorage_client: bool = True,
     ) -> "pyarrow.Table":
         """[Beta] Create a class:`pyarrow.Table` by loading all pages of a
@@ -1723,7 +1724,7 @@ def to_arrow(
 
     def to_dataframe_iterable(
         self,
-        bqstorage_client: Optional[bigquery_storage.BigQueryReadClient] = None,
+        bqstorage_client: Optional["bigquery_storage.BigQueryReadClient"] = None,
         dtypes: Dict[str, Any] = None,
         max_queue_size: int = _pandas_helpers._MAX_QUEUE_SIZE_DEFAULT,
     ) -> "pandas.DataFrame":
@@ -1797,7 +1798,7 @@ def to_dataframe_iterable(
     # changes to job.QueryJob.to_dataframe()
     def to_dataframe(
         self,
-        bqstorage_client: Optional[bigquery_storage.BigQueryReadClient] = None,
+        bqstorage_client: Optional["bigquery_storage.BigQueryReadClient"] = None,
         dtypes: Dict[str, Any] = None,
         progress_bar_type: str = None,
         create_bqstorage_client: bool = True,
@@ -1978,7 +1979,7 @@ def to_dataframe(
 
     def to_dataframe_iterable(
         self,
-        bqstorage_client: Optional[bigquery_storage.BigQueryReadClient] = None,
+        bqstorage_client: Optional["bigquery_storage.BigQueryReadClient"] = None,
         dtypes: Optional[Dict[str, Any]] = None,
         max_queue_size: Optional[int] = None,
     ) -> Iterator["pandas.DataFrame"]: