Peter Visscher and colleagues examine the genetic architecture of body mass index (BMI) and height using a within-family approach that uses realized genetic sharing between 20,240 sibling pairs from 9,577 nuclear families (Am. J. Hum. Genet. doi:10.1016/j.ajhg.2013.10.005, 31 October 2013). They calculate identity by descent (IBD) for each sibling pair using ∼20,000 SNPs in linkage equilibrium. They estimate the narrow-sense heritability for BMI at 0.42 (s.e.m. = 0.17) and for height at 0.69 (s.e.m. = 0.14), and suggest that previous heritability estimates for BMI were likely overestimates. They find a positive correlation of partitioned genetic variance with chromosome size for both BMI and height, consistent with a polygenic architecture. They perform a within-family linkage study for height and BMI and show increasing genomic inflation factors proportional to sample size. Their simulations show that this linkage analysis is robust to population stratification as well as to genetic heterogeneity between cohorts. They further demonstrate that the observed data are consistent with simulations assuming a polygenic architecture including both common and rare variation. They also suggest that polygenic architecture has contributed to failures to replicate linkage studies. They estimate that 67% and 40% of variance for height and BMI, respectively, is captured by common SNPs.