In this study we explore whether a phenocopy frequency (defined as a "penetrance' for nondisease genotypes) can approximate or model genetic heterogeneity in a single-locus analysis. We simulated two types of heterogeneity situations: "sporadic models', where there are two forms of a disease, one genetic and linked to a marker and the other purely random, and "genetic heterogeneity models', where the disease is caused by either of two different loci, one linked to the marker and the other unlinked. We analyzed simulated data sets for linkage, assuming a single-locus analysis with varying phenocopy frequency, in analogy with earlier work on epistatic two-locus models. We found that in the presence of purely random sporadics, there was a difference between assuming any nonzero phenocopy frequency and a zero frequency, but that the actual value of the assumed phenocopy frequency had little effect on the maximum lod score. In contrast, when both forms of disease are genetic, and are generated under similar genetic parameters, assuming a positive phenocopy frequency will not, in general, compensate for the presence of the unlinked form. However, when the modes of inheritance of the two forms differ, the assumption of a nonzero phenocopy frequency does have an effect, either to increase or decrease the maximum lod score, depending on the modes of inheritance of the two disease forms. We conclude with practical recommendations for investigators, based on these results.