import pandas as pd df = pd.DataFrame({'A':[1, 2, 3, 4, 5, 6], 'B':[0, 1, 2, 0, 1, 2]}) df.groupby('B').apply(lambda x: x.sort_values('A', ascending = True))