The mating system in an experimental population of rye (Secale cereale L.) cv. Merced was studied using eight polymorphic isozyme loci as genetic markers and two progeny samples from a large population: 20 emasculated mother plants (obligately cross-pollinating) and 100 non-emasculated mother plants. Merced rye showed a significant frequency of self-fertilization (s 1 -t) in a previous generation, but in the generation under study the outcrossing frequency was not significantly different from t 1. This fact suggests that environmental factors can affect the rate of outcrossing vs. selfing (self-fertilization). In spite of the outcrossing rate of t = 1, the large progenies from emasculated plants (100 seeds per plant) demonstrated that pollination was not completely at random. This represents a violation of the generally accepted mixed-mating model on which most methods used to estimate mating parameters are based. Temporal heterogeneity and the differential ability of gametophytes to fertilize are possible causes of the non-randomness of mating. It is concluded that, although violations of the assumption of the mixed-mating model can bias the estimates, most of the estimated outcrossing values clearly differ from complete outcrossing.