df.sort_values(by='col4', key=lambda col: col.str.lower()) col1 col2 col3 col4 0 A 2 0 a 1 A 1 1 B 2 B 9 9 c 3 NaN 8 4 D 4 D 7 2 e 5 C 4 3 F