Glasshouse and field experiments were conducted in 1993 at the International Livestock Research Institute (ILRI) in Addis Ababa to identify effective combinations of Trifolium semipilosum genotypes and Rhizobium leguminosarum bv. trifolii strains and to investigate the effects of waterlogging of Vertisols on the response of clover to inoculation. In the glasshouse, the Rhizobium strain ILCA372str produced nearly ten times as much nodule DM, three times as much shoot DM, five times as much root DM, and twice as much plant N on cv. Safari as it did on accession ILCA7609 under normal watering conditions. This spontaneous streptomycin-resistant mutant strain performed better than its parent strain. Shoot DM and N yields of this combination were, however, only 55 and 71%, respectively, of the corresponding N-fertilizer (150 kg N ha−1) treatment. Waterlogging reduced nodule DM by 33%, shoot DM by 41%, root DM by 25% and N yield by 33%, and responses of both plant genotypes to inoculation and to N fertilizer were reduced. In the field, there were no significant differences in nodulation, plant DM and plant N between seed and soil inoculation methods. Dry matter and N yields of Safari inoculated with strain ILCA372str were at least twice as high as those produced by any other strain or N fertilizer supplied at 100 kg N ha−1. Waterlogging during early establishment in the field probably reduced the availability of N fertilizer to the plants. Further field testing of the Safari-ILCA372str combination is recommended, particularly in combination with broad-bed and furrow (BBF) technology to drain Vertisols.