In this study we estimate the frequency at which P-element insertion events, as identified by in situ hybridization, generate lethal and mild viability mutations. The frequency of lethal mutations generated per insertion event was 0·004. Viability dropped an average of 1% per insertion event. Our results indicate that it is deletions and rearrangements resulting from the mobilization of P elements already in place and not the insertions per se that cause the drastic effects on viability and fitness observed in most studies of P–M dysgenesis-derived mutations. Elements of five other families (I, copia, 412, B104, and gypsy) were not mobilized in these crosses. Finally, we contrast the density of P elements on the X chromosome with the density on the four autosomal arms in a collection of thirty genomes from an African population. The relative number of P elements on the X chromosome is too high to be explained by either a hemizygous selection or a neutrality model. The possible reasons for the failure to detect selection are discussed.