[Truncated abstract] Evolution in two genetically diverse sown populations of subterranean clover (Trifolium subterraneum L.) was examined over 16 years at Mt Barker, a long growing season site, and at Nabawa, a short growing season site, in south-west Western Australia. One population consisted of a mixture of 40 strains sown in equal proportions, while the other was a bulk-hybrid population consisting of F2 seed from 253 crosses. Seed harvested annually and kept in cold storage was grown in an irrigated common garden at the University of Western Australia Field Station at Shenton Park, along with samples of the ancestral mixtures, to examine changes within these populations. Evolution in the strain mixture populations was measured by changes in strain frequency; strains were considered ecologically successful if their relative contribution was maintained or increased from that at sowing. Mixtures containing an additional 12 cultivars were also sown at each site to examine short-term population changes. The relative importance of 38 attributes measured in single-strain swards and spaced plants at Nabawa and Mt Barker was then related to strain success in mixtures at each site. Evolution in the bulk hybrid populations was measured in spaced plants at Shenton Park by changes in mean values and variability of 26 attributes. The ancestral populations evolved into markedly different populations at each site. Most evolution occurred within three years of sowing, due to elimination of poorly adapted genotypes. Within sites, the direction of selection was similar for both mixtures. In the strain mixture populations, different strains became dominant at each site. Divergent strains, those not identical to the sown strains, were significant components of the populations, particularly at Mt Barker, but their frequency did not increase. In the bulk hybrid populations, changes occurred in the means of 20 characters and variability declined in 11 characters at one or both sites. Appropriate flowering time was fundamental for success in both environments. At Nabawa, early flowering was crucial, while successful genotypes at Mt Barker were midseason and late flowering. Flowering time in both environments was a compromise between sufficient earliness for adequate seed production prior to the onset of summer drought, and deferment for as long as possible to allow plants to develop maximum size to compete for light in spring. The balance between these two competing forces differed in response to length of the growing season at each site