Previously we performed a 1012-generation mutation accumulation (MA) study in yeast and found that a surprisingly large proportion of fitness-altering mutations were beneficial. To verify this result and assess the impact of sampling error in our previous study, we have continued the MA experiment for an additional 1050 cell generations and re-estimated mutation parameters. After correcting for biases due to selection, we estimate that 13% of the mutations accumulated during this study are beneficial. We conclude that the high proportions of beneficial mutations observed in this and our previous study cannot be explained by sampling error. We also estimate the genome-wide mutation rate to be 13·7 × 10−5 mutations per haploid genome per cell generation and the absolute value of the average heterozygous effect of a mutation to be 7.3%.