Selecting rows from a Pandas dataframe with a compound (hierarchical) index

Try using xs to be very precise:

In [5]: df.xs('a', level=0)
Out[5]: 
        value1  value2
group2                
c          1.1     7.1
c          2.0     8.0
d          3.0     9.0

In [6]: df.xs('c', level="group2")
Out[6]: 
        value1  value2
group1                
a          1.1     7.1
a          2.0     8.0