To assemble mitotic spindles, cells nucleate microtubules from a variety of sources including chromosomes and centrosomes. We know little about how the regulation of microtubule nucleation contributes to spindle bipolarity and spindle size. The Aurora A kinase activator TPX2 is required for microtubule nucleation from chromosomes as well as for spindle bipolarity. We use bacterial artificial chromosome–based recombineering to introduce point mutants that block the interaction between TPX2 and Aurora A into human cells. TPX2 mutants have very short spindles but, surprisingly, are still bipolar and segregate chromosomes. Examination of microtubule nucleation during spindle assembly shows that microtubules fail to nucleate from chromosomes. Thus, chromosome nucleation is not essential for bipolarity during human cell mitosis when centrosomes are present. Rather, chromosome nucleation is involved in spindle pole separation and setting spindle length. A second Aurora A–independent function of TPX2 is required to bipolarize spindles.