Several methods for aggregate rare-variant association testing have recently been reported, including collapsing or weighting methods and gene- or region-based association tests. Although it is possible to estimate the average genetic effect for a group of rare variants from aggregate tests, there are potential biases, including winner's curse, selection procedures and differences between populations. Suzanne Leal and Dajiang Liu now report a new method to correct for bias in estimating the average genetic effect of a group of rare variants jointly analyzed for association consisting of a resampling-based approach and a bootstrap-sample-split algorithm (Am. J. Hum. Genet. 91, 585–596, 2012). They compare methods for estimating the average genetic effect and variance across a range of models in simulations, finding that the estimated variance is always less than the true locus-specific genetic variance, due to the inclusion of non-causal variants as well as causal variants with heterogeneous effects. The authors report the application of the new method to a resequencing data set of 4 genes in 1,045 individuals from the Dallas Heart Study, testing rare-variant associations with metabolic quantitative traits. The authors demonstrate the efficient estimation of average genetic effects in joint analysis of rare variants and note that estimated variance should be considered as a lower bound for the locus-specific variance.