HeLa cells blocked in metaphase with 0.04 micrograms/ml of the microtubule poison nocodazole were shown to contain large numbers of microtubules with typical mitotic organization but no cenriole. Lysis of nocodazole-poisoned cells in a microtubule reassembly buffer containing 0.5 M PIPES, 2.5% dimethyl sulfoxide, 1 mM EDTA, 1 mM MgCl2, 1 mM GTP, 1% Triton X-165, 0.5% sodium deoxycholate, 0.2% SDS, pH 6.9, preserved metaphase aster structures 5 micrograms in diameter surrounded only by a thin, fibrous cell remnant. Inclusion of 2 mg/ml porcine brain microtubule protein in the lysis buffer produced asters up to 20 micrometers in diameter with a birefringent retardation of 5-6 nm. In these large asters the central microtubules had normal morphology, but peripheral microtubules were clearly abnormal. Our interpretation is that in high PIPES lysis buffer, exogenous brain tubulin adds to the distal ends of preexisting aster microtubules to form abnormal microtubules. This observation supports the assumptions made by Borisy and by Summers and Kirschner in their interpretation of growth experiments to determine the microtubule polarity in mitotic structures.