Methods to determine the proportion of variability manifested by a quantitative trait that is attributable to additive genetic effects are important tools for human population and statistical geneticists. Though traditional methods based on parent-offspring and sib-pair correlations have been well researched, they are being steadily supplanted or complemented by more powerful pedigree-based variance-component techniques. In this paper, a theoretical investigation of the relative efficiency and power of small (i.e., < 10 members) variance-component pedigree designs for heritability estimation is undertaken. The information gain in adding sibs and generations to pedigrees is discussed. Sample size guidelines based on theoretical power functions are offered, as are directions for future research.