A gradient of soils with increasing metal contamination was constructed by mixing uncontaminated farmyard manure-treated (FYM) and contaminated, sewage sludge-treated (S) soils from the Woburn Experimental Farm, and separate samples of these soils were inoculated with strains of Rhizobium leguminosarum bv. trifolii, R. meliloti and R. loti. The survival of rhizobia in these soils was monitored using most-probable number (MPN) plant infection tests. After 51 days incubation numbers of R. meliloti were similar in all of the soils irrespective of the heavy metal content, whilst the numbers of R. leguminosarum bv. trifolii and R. loti were reduced in soils containing more than 2 3 sludge-treated soil. The numbers of R. leguminosarum bv. trifolii were monitored in the same gradient of soils over a 171 day period and differences in survival between the soils were only apparent from 27 days in soil mixtures containing more than 5 6 sludge-treated soil. After 171 days the number of cells surviving had decreased even in the soil mixture containing 1 6 sludge-treated soil when compared with the uncontaminated FYM-treated soil. The populations of indigenous R. leguminosarum bv. trifolii were also monitored in the soil of the FYM-treated and S-treated plots in the field. Numbers of rhizobia remained at around 4 × 10 4 cells g soil −1 in the FYM-treated soil over 3 yr after clover was removed, but gradually declined in the S-treated soil to 1 × 10 3 cells g soil −1.