Three hybridoma clones, which were shown to change the characteristics of their antibody specificities when grown under different culturing conditions, are described in detail. This phenomenon was shown to be due to the persistence of mixed clones, even under conditions where standard statistical treatment indicated a high probability of monoclonality. Such mixed clones persisted, sometimes undetected, through repeated cycles of re-cloning. It was shown that the assumption that every viable clone has the same random chance of monoclonality, is invalid, and can lead to misleadingly high estimates for the probability of monoclonality. Verification of seeding of individual wells with single cells is recommended and the relative merits of this versus repeated limiting-dilution cloning are discussed.