diff --git a/bigframes/core/blocks.py b/bigframes/core/blocks.py
index 0f98f582c2..ca06c8859e 100644
--- a/bigframes/core/blocks.py
+++ b/bigframes/core/blocks.py
@@ -3432,7 +3432,6 @@ def unpivot(
             array_value, type="cross"
         )
     new_passthrough_cols = [column_mapping[col] for col in passthrough_columns]
-    # Last column is offsets
     index_col_ids = [labels_mapping[col] for col in labels_array.column_ids[:-1]]
     explode_offsets_id = labels_mapping[labels_array.column_ids[-1]]
 
@@ -3442,6 +3441,10 @@ def unpivot(
     for input_ids in unpivot_columns:
         # row explode offset used to choose the input column
         # we use offset instead of label as labels are not necessarily unique
+        if not input_ids:
+            unpivot_exprs.append(ex.const(None))
+            continue
+
         cases = itertools.chain(
             *(
                 (
@@ -3459,7 +3462,7 @@ def unpivot(
     joined_array, unpivot_col_ids = joined_array.compute_values(unpivot_exprs)
 
     return joined_array.select_columns(
-        [*index_col_ids, *unpivot_col_ids, *new_passthrough_cols]
+        [*index_col_ids, *unpivot_col_ids, *new_passthrough_cols], allow_renames=True
     ), (tuple(index_col_ids), tuple(unpivot_col_ids), tuple(new_passthrough_cols))
 
 
@@ -3471,18 +3474,31 @@ def _pd_index_to_array_value(
     Create an ArrayValue from a list of label tuples.
     The last column will be row offsets.
     """
+    id_gen = bigframes.core.identifiers.standard_id_strings()
+    index_ids = [next(id_gen) for _ in range(index.nlevels)]
+    offset_id = next(id_gen)
+
     rows = []
     labels_as_tuples = utils.index_as_tuples(index)
     for row_offset in range(len(index)):
-        id_gen = bigframes.core.identifiers.standard_id_strings()
         row_label = labels_as_tuples[row_offset]
-        row_label = (row_label,) if not isinstance(row_label, tuple) else row_label
-        row = {}
-        for label_part, id in zip(row_label, id_gen):
-            row[id] = label_part if pd.notnull(label_part) else None
-        row[next(id_gen)] = row_offset
+        row = {
+            id: (val if pd.notnull(val) else None)
+            for id, val in zip(index_ids, row_label)
+        }
+        row[offset_id] = row_offset
         rows.append(row)
 
+    if not rows:
+        # Create empty table with correct columns
+        schema = pa.schema(
+            [pa.field(id, pa.null()) for id in index_ids]
+            + [pa.field(offset_id, pa.int64())]
+        )
+        return core.ArrayValue.from_pyarrow(
+            pa.Table.from_batches([], schema=schema), session=session
+        )
+
     return core.ArrayValue.from_pyarrow(pa.Table.from_pylist(rows), session=session)
 
 
diff --git a/bigframes/pandas/core/methods/describe.py b/bigframes/pandas/core/methods/describe.py
index 6fd7960daf..eb12ac7831 100644
--- a/bigframes/pandas/core/methods/describe.py
+++ b/bigframes/pandas/core/methods/describe.py
@@ -17,6 +17,7 @@
 import typing
 
 import pandas as pd
+import pyarrow as pa
 
 from bigframes import dataframe, dtypes, series
 from bigframes.core import agg_expressions, blocks
@@ -86,9 +87,13 @@ def _describe(
         if include != "all" and dtype not in _DEFAULT_DTYPES:
             continue
         agg_ops = _get_aggs_for_dtype(dtype)
-        stats.extend(op.as_expr(col_id) for op in agg_ops)
-        label_tuple = (label,) if block.column_labels.nlevels == 1 else label
-        column_labels.extend((*label_tuple, op.name) for op in agg_ops)  # type: ignore
+
+        label_tuple = (
+            (label,) if block.column_labels.nlevels == 1 else typing.cast(tuple, label)
+        )
+        for op in agg_ops:
+            stats.append(op.as_expr(col_id))
+            column_labels.append((*label_tuple, op.name))
 
     agg_block = block.aggregate(
         by_column_ids=by_col_ids,
@@ -100,7 +105,7 @@ def _describe(
 
 
 def _get_aggs_for_dtype(dtype) -> list[aggregations.UnaryAggregateOp]:
-    if dtype in dtypes.NUMERIC_BIGFRAMES_TYPES_RESTRICTIVE:
+    if dtypes.is_numeric(dtype, include_bool=False):
         return [
             aggregations.count_op,
             aggregations.mean_op,
@@ -111,14 +116,18 @@ def _get_aggs_for_dtype(dtype) -> list[aggregations.UnaryAggregateOp]:
             aggregations.ApproxQuartilesOp(3),
             aggregations.max_op,
         ]
-    elif dtype in dtypes.TEMPORAL_NUMERIC_BIGFRAMES_TYPES:
+    elif dtypes.is_datetime_like(dtype) or dtypes.is_date_like(dtype):
         return [aggregations.count_op]
-    elif dtype in [
-        dtypes.STRING_DTYPE,
-        dtypes.BOOL_DTYPE,
-        dtypes.BYTES_DTYPE,
-        dtypes.TIME_DTYPE,
-    ]:
+    elif (
+        dtypes.is_string_like(dtype)
+        or dtypes.is_binary_like(dtype)
+        or dtypes.is_time_like(dtype)
+        or (
+            isinstance(dtype, pd.ArrowDtype)
+            and pa.types.is_struct(dtype.pyarrow_dtype)
+            and not dtypes.contains_db_dtypes_json_dtype(dtype)
+        )
+    ):
         return [aggregations.count_op, aggregations.nunique_op]
     else:
-        return []
+        return [aggregations.count_op]
diff --git a/tests/unit/core/test_blocks_unpivot.py b/tests/unit/core/test_blocks_unpivot.py
new file mode 100644
index 0000000000..b1bde67ff3
--- /dev/null
+++ b/tests/unit/core/test_blocks_unpivot.py
@@ -0,0 +1,79 @@
+# Copyright 2025 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from unittest import mock
+
+import pandas as pd
+import pytest
+
+from bigframes.core import blocks
+
+
+@pytest.fixture
+def mock_session():
+    session = mock.MagicMock()
+    session.bqclient = None
+    return session
+
+
+def test_pd_index_to_array_value_with_empty_index_creates_no_columns(mock_session):
+    """
+    Tests that `_pd_index_to_array_value` with an empty pandas Index creates
+    an ArrayValue with the expected number of columns (index level + offset).
+    """
+    empty_index = pd.Index([], name="test")
+
+    array_val = blocks._pd_index_to_array_value(mock_session, empty_index)
+
+    # 1 index level + 1 offset column
+    assert len(array_val.column_ids) == 2
+
+
+def test_pd_index_to_array_value_with_empty_multiindex_creates_no_columns(mock_session):
+    """
+    Tests that `_pd_index_to_array_value` with an empty pandas MultiIndex creates
+    an ArrayValue with the expected number of columns (index levels + offset).
+    """
+    empty_index = pd.MultiIndex.from_arrays([[], []], names=["a", "b"])
+
+    array_val = blocks._pd_index_to_array_value(mock_session, empty_index)
+
+    # 2 index levels + 1 offset column
+    assert len(array_val.column_ids) == 3
+
+
+def test_unpivot_with_empty_row_labels(mock_session):
+    """
+    Tests that `unpivot` handles an empty `row_labels` index correctly by producing 0 rows.
+    """
+    import pyarrow as pa
+
+    # Create a dummy ArrayValue
+    df = pd.DataFrame({"a": [1, 2, 3], "b": [4, 5, 6]})
+    pa_table = pa.Table.from_pandas(df)
+    array_value = blocks.core.ArrayValue.from_pyarrow(pa_table, session=mock_session)
+
+    # Call unpivot with an empty pd.Index
+    unpivot_result, (index_cols, value_cols, passthrough_cols) = blocks.unpivot(
+        array_value,
+        row_labels=pd.Index([]),
+        unpivot_columns=[("a",)],
+        passthrough_columns=["b"],
+    )
+
+    # The expected behavior is that the unpivot operation produces 0 rows.
+    assert unpivot_result is not array_value
+    assert index_cols == ("col_0",)
+    assert len(value_cols) == 1
+    assert passthrough_cols == ("b",)