Community matrix theory has been proposed as a means of predicting whether a particular set of species will form a stable mixture. However, the approach has rarely been used with data from real communities. Using plant competition experiments, we use community matrix theory to predict the stability and competitive structuring of a lawn community. Seven species from the lawn, including the six most abundant, were grown in boxes, in conditions very similar to those on the lawn. They were grown alone (monocultures), and in all possible pairs. The species formed a transitive hierarchy of competitive ability, with most pairs of species showing asymmetric competition. Relative competitive ability (competitive effect) was positively correlated with published estimates of the maximum relative growth rate (RGRmax) for the same species. A seven-species community matrix predicted the mixture of species to be unstable. Simulations revealed two topological features of this community matrix. First, the matrix was closer to the stability/instability boundary than predicted from a range of null (random) models, suggesting that the lawn may be close to stability. Second, the tendencies of the lawn species to compete asymmetrically, and to be arranged in competitive hierarchies, were found to be positively associated with stability, and hence may be contributing factors to the near-stability seen in the matrix. The limitations of using competition experiments for constructing community matrices are discussed.