Genotype error can greatly reduce the power of a genetic study. For family data, genotype error can be assessed by examining marker data for non-Mendelian inconsistencies, closely linked markers for double recombination events, and consistency of duplicate genotypes. For case-control data, duplicate samples are genotyped, and controls are tested for deviations from Hardy-Weinberg equilibrium (HWE). Duplicate samples can provide accurate estimates of genotyping error rates, unless systematic genotyping errors have occurred. Although genotyping errors can cause deviations from HWE, these deviations are usually small, and the power to detect them is low except for high rates of genotyping error and/or large sample sizes. An additional problem is that even when deviations from HWE are detected for marker loci, without additional experimentation it is not possible to unequivocally implicate genotyping error as the cause. The power and sample sizes necessary to detect deviations from HWE for single-nucleotide polymorphism (SNP) data are examined for a variety of genotyping error and pseudo-SNP models. For the majority of genotyping models examined, the power is poor to detect deviations from HWE. For example, for 1,000 controls, if an allele with a frequency of 0.1 fails to amplify for 28% of the heterozygous genotypes producing a sample error rate of 0.05, the power is 0.51 to detect a deviation from HWE at an alpha level of 0.05. On the other hand, the detection of deviations from HWE for pseudo-SNPs (paralogous and ectopic sequence variants) for the majority of models examined produces a power of >0.8 for sample sizes as small as 50 individuals. Genet. Epidemiol. © 2005 Wiley-Liss, Inc.