diff --git a/python/cudf_polars/cudf_polars/dsl/expr.py b/python/cudf_polars/cudf_polars/dsl/expr.py
index 871134665af..99e5a59de2f 100644
--- a/python/cudf_polars/cudf_polars/dsl/expr.py
+++ b/python/cudf_polars/cudf_polars/dsl/expr.py
@@ -27,11 +27,12 @@
 import cudf._lib.pylibcudf as plc
 
 from cudf_polars.containers import Column, NamedColumn
-from cudf_polars.utils import sorting
+from cudf_polars.utils import dtypes, sorting
 
 if TYPE_CHECKING:
     from collections.abc import Mapping, Sequence
 
+    import polars.polars as plrs
     import polars.type_aliases as pl_types
 
     from cudf_polars.containers import DataFrame
@@ -368,6 +369,29 @@ def do_evaluate(
         return Column(plc.Column.from_scalar(plc.interop.from_arrow(self.value), 1))
 
 
+class LiteralColumn(Expr):
+    __slots__ = ("value",)
+    _non_child = ("dtype", "value")
+    value: pa.Array[Any, Any]
+    children: tuple[()]
+
+    def __init__(self, dtype: plc.DataType, value: plrs.PySeries) -> None:
+        super().__init__(dtype)
+        data = value.to_arrow()
+        self.value = data.cast(dtypes.downcast_arrow_lists(data.type))
+
+    def do_evaluate(
+        self,
+        df: DataFrame,
+        *,
+        context: ExecutionContext = ExecutionContext.FRAME,
+        mapping: Mapping[Expr, Column] | None = None,
+    ) -> Column:
+        """Evaluate this expression given a dataframe for context."""
+        # datatype of pyarrow array is correct by construction.
+        return Column(plc.interop.from_arrow(self.value))
+
+
 class Col(Expr):
     __slots__ = ("name",)
     _non_child = ("dtype", "name")
diff --git a/python/cudf_polars/cudf_polars/dsl/translate.py b/python/cudf_polars/cudf_polars/dsl/translate.py
index 5d289885f47..11c6f8301f7 100644
--- a/python/cudf_polars/cudf_polars/dsl/translate.py
+++ b/python/cudf_polars/cudf_polars/dsl/translate.py
@@ -12,6 +12,7 @@
 import pyarrow as pa
 from typing_extensions import assert_never
 
+import polars.polars as plrs
 from polars.polars import _expr_nodes as pl_expr, _ir_nodes as pl_ir
 
 import cudf._lib.pylibcudf as plc
@@ -373,6 +374,8 @@ def _(node: pl_expr.Window, visitor: NodeTraverser, dtype: plc.DataType) -> expr
 
 @_translate_expr.register
 def _(node: pl_expr.Literal, visitor: NodeTraverser, dtype: plc.DataType) -> expr.Expr:
+    if isinstance(node.value, plrs.PySeries):
+        return expr.LiteralColumn(dtype, node.value)
     value = pa.scalar(node.value, type=plc.interop.to_arrow(dtype))
     return expr.Literal(dtype, value)
 
diff --git a/python/cudf_polars/tests/expressions/test_series_literal.py b/python/cudf_polars/tests/expressions/test_series_literal.py
new file mode 100644
index 00000000000..c7613cf1886
--- /dev/null
+++ b/python/cudf_polars/tests/expressions/test_series_literal.py
@@ -0,0 +1,19 @@
+# SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES.
+# SPDX-License-Identifier: Apache-2.0
+from __future__ import annotations
+
+import polars as pl
+
+from cudf_polars.testing.asserts import assert_gpu_result_equal
+
+
+def test_select_series_literal():
+    df = pl.LazyFrame({})
+
+    q = df.select(
+        a=pl.Series([1, 2, 3]),
+        b=pl.Series(["a", "b", "c"], dtype=pl.String()),
+        c=pl.Series([[1, 2], [3], None], dtype=pl.List(pl.UInt16())),
+    )
+
+    assert_gpu_result_equal(q)