diff --git a/python/lsst/daf/butler/formatters/parquet.py b/python/lsst/daf/butler/formatters/parquet.py
index 5c9c952bf4..6e8937a019 100644
--- a/python/lsst/daf/butler/formatters/parquet.py
+++ b/python/lsst/daf/butler/formatters/parquet.py
@@ -1098,7 +1098,8 @@ def _arrow_string_to_numpy_dtype(
         # String/bytes length from header.
         strlen = int(schema.metadata[encoded])
     elif numpy_column is not None and len(numpy_column) > 0:
-        strlen = max([len(row) for row in numpy_column if row])
+        lengths = [len(row) for row in numpy_column if row]
+        strlen = max(lengths) if lengths else 0
 
     dtype = f"U{strlen}" if schema.field(name).type == pa.string() else f"|S{strlen}"