SUMMARYThe polygenic threshold model assumes that the distribution of the underlying liability is approximately normal. This paper examines the impact of deviations from normality in the underlying liability distribution when the number of loci affecting liability is finite, but large (e.g. 5–10). Skewness and kurtosis of the liability distribution are produced by varying the frequency of pathogenic alleles while population risk of illness is held constant by parallel changes in the threshold of manifestation. For a given population risk of illness, the recurrence risk in relatives varies widely as a function of the shape of the liability distribution. These results raise questions about the accuracy of the polygenic threshold approximation for the calculation of recurrence risks for human ‘polygenic’ diso