Just promoting @felix-benning's amazing tip to a full answer – set include_groups=False, then use df.name to access the group name. NB I only tested this with a single item in groupby, but it sounds like it should work with multiple ones.