Repeated efforts to estimate the genomic deleterious mutation rate per generation (U) in Drosophila melanogaster have yielded inconsistent estimates ranging from 0·01 to nearly 1. We carried out a mutation-accumulation experiment with a cryopreserved control population in hopes of resolving some of the uncertainties raised by these estimates. Mutation accumulation (MA) was carried out by brother–sister mating of 150 sublines derived from two inbred lines. Fitness was measured under conditions chosen to mimic the ancestral laboratory environment of these genotypes. We monitored the insertions of a transposable element, copia, that proved to accumulate at the unusually high rate of 0·24 per genome per generation in one of our MA lines. Mutational variance in fitness increased at a rate consistent with previous studies, yielding a mutational coefficient of variation greater than 3%. The performance of the cryopreserved control relative to the MA lines was inconsistent, so estimates of mutation rate by the Bateman–Mukai method are suspect. Taken at face value, these data suggest a modest decline in fitness of about 0·3% per generation. The element number of copia was a significant predictor of fitness within generations; on average, insertions caused a 0·76% loss in fitness, although the confidence limits on this estimate are wide.