Surprised nobody brought this one up:
# To remove last n rows
df.head(-n)
# To remove first n rows
df.tail(-n)
Running a speed test on a DataFrame of 1000 rows shows that slicing and head/tail are ~6 times faster than using drop:
>>> %timeit df[:-1]
125 µs ± 132 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)
>>> %timeit df.head(-1)
129 µs ± 1.18 µs per loop (mean ± std. dev. of 7 runs, 10000 loops each)
>>> %timeit df.drop(df.tail(1).index)
751 µs ± 20.4 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)