Quantitative analyses of fast- and slow-growing soybean rhizobia populations in soils of four different provinces of China (Hubei, Shan Dong, Henan, and Xinjiang) have been carried out using the most probable number technique (MPN). All soils contained fast- (FSR) and slow-growing (SSR) soybean rhizobia. Asiatic and American soybean cultivars grown at acid, neutral and alkaline pH were used as trapping hosts for FSR and SSR strains. The estimated total indigenous soybean-rhizobia populations of the Xinjiang and Shan Dong soil samples greatly varied with the different soybean cultivars used. The soybean cultivar and the pH at which plants were grown also showed clear effects on the FSR/SSR rations isolated from nodules. Results of competition experiments between FSR and SSR strains supported the importance of the soybean cultivar and the pH on the outcome of competition for nodulation between FSR and SSR strains. In general, nodule occupancy by FSRs significantly increased at alkaline pH. Bacterial isolates from soybean cultivar Jing Dou 19 inoculated with Xinjiang soil nodulate cultivars Heinong 33 and Williams very poorly. Plasmid and lipopolysaccharide (LPS) profiles and PCR-RAPD analyses showed that cultivar Jing Dou 19 had trapped a diversity of FSR strains. Most of the isolates from soybean cultivar Heinong 33 inoculated with Xinjiang soil were able to nodulate Heinong 33 and Williams showed very similar, or identical, plasmid, LPS and PCR-RAPD profiles. All the strains isolated from Xinjiang province, regardless of the soybean cultivar used for trapping, showed similar nodulation factor (LCO) profiles as judged by thin layer chromatographic analyses. These results indicate that the existence of soybean rhizobia sub-populations showing marked cultivar specificity, can affect the estimation of total soybean rhizobia populations indigenous to the soil, and can also affect the diversity of soybean rhizobial strains isolated from soybean nodules.