Microbial degradation of 3-(3,4-dichlorophenyl)-1,1-dimethylurea (diuron) and mineralization of 4-chloro-2-methylphenoxyacetic acid (MCPA) were studied in soil samples taken from the ballast layers of three Swedish railway embankments. The degradation of diuron followed first-order kinetics and half-lives ranged between 122-365 days. The halflives correlated strongly with microbial biomass estimated by substrate-induced respiration (SIR; R = -0.85; p<0.05) and with the amount of organic matter measured as loss on ignition (R = -0.87; p<0.05). Accumulation of the metabolites DCPMU and DCPU was observed in all samples and these were only detectably degraded in the sample with the highest SIR. Addition of ground lucerne straw to the ballast samples stimulated microbial activity and led to increased formation of metabolites, but further transformation of DCPMU and DCPU was not enhanced. Mineralization of MCPA followed growth-linked kinetics and the time for 50% mineralization was 44.5 ± 7.1 days in samples of previously untreated ballast. In samples of ballast that had been previously treated with the herbicide formulation MCPA 750, the time for 50% mineralization was reduced to 13.7 ±11.3 days. The number of MCPA degraders, quantified using an MPN technique, was clearly increased but highly variable. An average yield of 0.18 cells pg -1 of MCPA was estimated from the kinetic data. The yield estimates correlated with the amount of nitrogen in the ballast, indicating that mineralization of MCPA was nitrogen-limited in the railway embankments studied. This has practical implications for weed control using herbicides on railways.