Genotypic errors, whether due to mutation or laboratory error, can cause the genotypes of parents and their offspring to appear inconsistent with Mendelian inheritance. As a result, molecular parentage analyses are expected to benefit when allowances are made for the presence of genotypic errors. However, a cost of allowing for genotypic errors might also be expected under some analytical conditions, primarily because parentage analyses that assume nonzero genotypic error rates can neither assign nor exclude parentage with certainty. The goal of this work was therefore to determine whether or not such costs might be important under conditions relevant to parentage analyses, particularly in natural populations. Simulation results indicate that the costs may often outweigh the benefits of accounting for nonzero error rates, except in situations where data are available for many marker loci. Consequently, the most powerful approach to handling genotypic errors in parentage analyses might be to apply likelihood equations with error rates set to values substantially lower than the rates at which genotypic errors occur. When applying molecular parentage analyses to natural populations, we advocate an increased consideration of optimal strategies for handling genotypic errors. Currently available software packages contain procedures that can be used for this purpose.