diff --git a/python/deltalake/table.py b/python/deltalake/table.py
index 6f335feba0..c7c5532eaa 100644
--- a/python/deltalake/table.py
+++ b/python/deltalake/table.py
@@ -37,8 +37,6 @@
 if TYPE_CHECKING:
     import os
 
-    import pandas
-
 from deltalake._internal import DeltaDataChecker as _DeltaDataChecker
 from deltalake._internal import RawDeltaTable
 from deltalake._internal import create_deltalake as _create_deltalake
@@ -48,6 +46,13 @@
 from deltalake.fs import DeltaStorageHandler
 from deltalake.schema import Schema as DeltaSchema
 
+try:
+    import pandas as pd  # noqa: F811
+except ModuleNotFoundError:
+    _has_pandas = False
+else:
+    _has_pandas = True
+
 MAX_SUPPORTED_READER_VERSION = 1
 MAX_SUPPORTED_WRITER_VERSION = 2
 
@@ -892,7 +897,7 @@ def merge(
             pyarrow.RecordBatch,
             pyarrow.RecordBatchReader,
             ds.Dataset,
-            "pandas.DataFrame",
+            "pd.DataFrame",
         ],
         predicate: str,
         source_alias: Optional[str] = None,
@@ -937,7 +942,7 @@ def merge(
             source = convert_pyarrow_table(source, large_dtypes)
         elif isinstance(source, ds.Dataset):
             source = convert_pyarrow_dataset(source, large_dtypes)
-        elif isinstance(source, pandas.DataFrame):
+        elif _has_pandas and isinstance(source, pd.DataFrame):
             source = convert_pyarrow_table(
                 pyarrow.Table.from_pandas(source), large_dtypes
             )
@@ -1094,7 +1099,7 @@ def to_pandas(
         columns: Optional[List[str]] = None,
         filesystem: Optional[Union[str, pa_fs.FileSystem]] = None,
         filters: Optional[FilterType] = None,
-    ) -> "pandas.DataFrame":
+    ) -> "pd.DataFrame":
         """
         Build a pandas dataframe using data from the DeltaTable.