The generation and analysis of mutants is central to studies of gene function in model organisms. Methods for random mutagenesis in Drosophila melanogaster have been available for many years, but an alternative approach--targeted mutagenesis using homologous recombination--has only recently been developed. This approach has the advantage of specificity, because genes of interest can be altered. One might expect with a gene-targeting approach that the frequency of background mutations would be minimal. Unfortunately, we have found that this is not the case. Although the possibility of background mutations arising during homologous-recombination-based gene targeting has been raised in the literature, it is not routinely taken into account when using this technique. Our experience suggests that it can be a considerable problem but that it has a relatively simple solution.