The factors that cause new mutations or affect the rate at which they occur have important implications for many areas of genetics. But recent work on phenomena such as premeiotic mutations, which yield a cluster of identical new mutants at the some time, led us to realize that researchers are using the term "mutation rate" in different, and sometimes contradictory, ways. One premeiotic genetic change may ultimately yield several new mutant offspring, but should this be considered one new mutation or many? The way the data are handled in analyses can have a significant effect on the results. How, then, does one handle clusters in the estimation of mutation rates? We explore this question and propose that geneticists begin to distinguish clearly between three different phenomena that to this point have been given the same name: the initial prerepair "genetic damage rate," the postrepair "mutational event rate," and the observed "mutation rate" as it is expressed in the proportion of new mutant offspring. We believe that all new mutant offspring should be counted when estimating mutation rate, irrespective of when in the developmental cycle it is believed that the initial mutational event occurred.