Although considerable evidence exists that plant competition is generally asymmetric or "one-sided", with larger plants having a disproportionate competitive effect on smaller plants, currently employed measures of local interference generally assume that competition is "two-sided". We describe a simple measure of competitive asymmetry in which the effects of neighbors smaller than a focal individual are discounted by a constant factor, and include this variable in a composite measure of local interference. In this model competition varies between complete asymmetry (the effects of smaller plants are entirely discounted) and complete symmetry (the competitive effect of a neighbor is proportional to its size). The proposed method is applied to two natural populations and one experimental monoculture. In all cases an asymmetric model provides the best fit to the data. Completely two-sided models account for 26-39% of the variance in relative growth rate, while relatively one-sided models account for 44-57%. The increases in r² values resulting from the inclusion of asymmetry are significant in the two cases in which the data permit randomization tests. Our results suggest that interference is completely asymmetric in a population of Impatiens pallida, a species with very low root allocation and a shallow crown, and somewhat less asymmetric in an experimental monoculture of Ambrosia artemisiifolia and a natural stand of Pinus rigida, cases in which competition for water and nutrient resources is likely to be of greater importance.