The relative nodulating success of strains of Rhizobium meliloti, with Medicago sativa and M. truncatula, and R. trifolii, with Trifolium repens, T. glomeratum and T. subterraneum, has been determined with inocula of paired competitors, supplied in various proportions. Host plants were raised from surface-sterilized seed, grown on agar in enclosed tubes, and populations and strain ratios were determined on the root surface as well as in the supplied inoculum. The use of heavy inoculum did not prevent all growth, and strain representation on the root surface could differ from that of the inoculum, because of either different adherence or the narrowing of extreme ratios with time. The ratios of strains obtained from nodules when related to those on the root surface yielded a 'competitive index', defined as the ratio of nodules due to each strain under conditions of equal representation on the root. In the five available cases inferior nitrogen-fixing effectiveness of a strain with a particular host was reflected in poor nodulating competitiveness, but similar differences were found in six comparisons between equally effective pairs, and with the one ineffective pair. Competitiveness did not correlate with the relative speed with which the strains produced nodules when used as a single culture or with the number of nodules they produced. Cases of mixed infections were encountered throughout the study; they generally accounted for less than 10% of the nodules typed, but in some combinations of host and strains the figure exceeded 25%.