The dissemination of catabolic plasmids was compared to bioaugmentation by strain inoculation in microcosm experiments. When Rhodococcus erythropolis strain T902, bearing a plasmid with trichloroethene and isopropylbenzene degradation pathways, was used as the inoculum, no transconjugant was isolated but the strain remained in the soil. This plasmid had a narrow host range. Pseudomonas putida strain C8S3 was used as the inoculum in a second approach. It bore a broad host range conjugative plasmid harboring a natural transposon, RP4::Tn4371, responsible for biphenyl and 4-chlorobiphenyl degradation pathways. The inoculating population slowly decreased from its original level (10(6) colony-forming units [CFU]/g of dry soil) to approx 3 x 10(2) CFU/g of dry soil after 3 wk. Transconjugant populations degrading biphenyl appeared in constant humidity soil (up to 2 x 10(3) CFU/g) and desiccating soil (up to 10(4) CFU/g). The feasibility of plasmid dissemination as a bioaugmentation technique was demonstrated in desiccating soils. The ecologic significance of desiccation in bioaugmentation was demonstrated: it upset the microbial ecology and the development of transconjugants.