With individual genes being identified that have an important effect on performance and fitness in livestock it is likely that such genes will be included in selection programmes. However, in order to devise sensible strategies to achieve this, knowledge of the frequency of the gene of interest is required. In practice, it is possible that estimates for gene frequency are based on genotype testing of only a subset of the population. The question then arises as to what conclusions can be drawn about the population gene frequency particularly in the likely scenario where the sample genotyped is not chosen at random. A procedure was developed by Van Arendonk et al. (1989) to have genotype information on as many individuals as possible given financial limits in the numbers genotyped. In this procedure the genotypes of individuals are predicted using pedigree information and the rules of Mendelian inheritance. What is less clear is the value of the additional information in predicting the population gene frequency. The objective of this study was to assess to consequence of sampling procedures on estimates of gene frequency when additional genotype information is or is not obtained by predicting genotypes on individuals themselves not tested.