The reproduction of spindle poles is a key event in the cell's preparation for mitosis. To gain further insight into how this process is controlled, we systematically characterized the ultrastructure of spindle poles whose reproductive capacity had been experimentally altered. In particular, we wanted to determine if the ability of a pole to reproduce before the next division is related to the number of centrioles it contains. We used mercaptoethanol to indirectly induce the formation of monopolar spindles in sea urchin eggs. We followed individually treated eggs in vivo with a polarizing microscope during the induction and development of monopolar spindles. We then fixed each egg at one of three predetermined key stages and serially semithick sectioned it for observation in a high-voltage electron microscope. We thus know the history of each egg before fixation and, from earlier studies, what that cell would have done had it not been fixed. We found that spindle poles that would have given rise to monopolar spindles at the next mitosis have only one centriole whereas spindle poles that would have formed bipolar spindles at the next division have two centrioles. By serially sectioning each egg, we were able to count all centrioles present. In the twelve cells examined, we found no cases of acentriolar spindle poles or centriole reduplication. Thus, the reproductive capacity of a spindle pole is linked to the number of centrioles it contains. Our experimental results also show, contrary to existing reports, that the daughter centriole of a centrosome can acquire pericentriolar material without first becoming a parent. Furthermore, our results demonstrate that the splitting apart of mother and daughter centrioles is an event that is distinct from, and not dependent on, centriole duplication.