Many plant species have mating systems characterized by a mixture of self-fertilization and outcrossing. Statistical estimation of the outcrossing rate has relied on a model of the mating process that assumes that successive outcross events within a family arise from independent draws of pollen from the total population of male plants. Although this assumption is likely to be most appropriate for wind-pollinated plants, it is not appropriate in certain insect-pollinated plants. An alternative model is developed that assumes that successive outcross events within a family involve pollen drawn front a single male parent. The estimation of the parameters that index this model is outlined and a procedure for calculating the variances of the parameter estimates is presented. Monte Carlo simulations of the sampling processes assumed by each model are also presented. The simulations show that application of the incorrect estimation model to data can lead to a large bias in parameter estimates.The mating system of a population determines how genetic information is transferred from one generation to the next (1). Moreover, the recombinational potential of the population may be strongly influenced by the mating system (2). Accurate information on the mating system in plant populations cannot be obtained through direct observation. This is due, in part, to difficulties inherent in tracing the flow of pollen among and within individuals. Plant population biologists have instead relied on methods to infer the mating system retrospectively from progeny genotypes. These methods use marker loci and statistical estimation (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18). A common practice in mating system studies that use allozymic markers is to assay several progeny from single maternal (seed) parents and estimate the maternal parent frequencies from the progeny genotype distributions using, as a basis, some underlying statistical model of the mating process. This logistically simple method of sampling the population avoids problems associated with age-and/or tissue-specific expression of allozyme phenotypes.In plant population genetics, one model of mating, the mixed-mating model, has held a central position in both theoretical and experimental investigations (7, 12). There are several reasons for choosing the mixed-mating model rather than assuming random mating to describe the pattern of gene transmission in plant populations. First, inbreeding is common in many plant species and the random-mating model is, therefore, inappropriate. Second, the mixed-mating model is simple and requires the estimation of only one additional parameter, the outcrossing rate. Third, the assumptions of the mixed-mating model are reasonable descriptions of the reproductive biology of many species. Despite the wide utility of the mixed-mating model, it may fail to properly describe the reproductive biology of certain insect-pollinated plants.Below we document this failure and then propose an alternative to the mixed-mating model th...