14C-ring-labelled isoproturon mineralisation was investigated in a French agricultural soil previously exposed to isoproturon. 50 different soil samples collected every 2 m along a transect of 100 m in length were treated one or two times with isoproturon under laboratory conditions and analysed by radiorespirometry. 94% of the soil samples showed a high ability to mineralise isoproturon with a relatively low variability in the cumulative percentage of mineralisation ranging from 30 to 51% of the initially added radioactivity for the samples treated once with the herbicide. About 45 to 67% of the initially added radioactivity was transformed into 14CO2 in soil samples treated twice with isoproturon. Fifty-one to 30% of the radioactive pesticide formed bound residues 120 days after the first isoproturon treatment. The radioactive compound incorporated in the microbial biomass ranged from 3 to 4% of the initially added radioactivity. The methanol-extractable radioactivity was negligible and in the majority of soil samples no metabolites could be detected by high performance liquid chromatography analyses. However, in one soil sample showing low ability to mineralise isoproturon, the mono-demethyl isoproturon derivative represented about 12% of the methanol-extractable fraction. After the second isoproturon treatment, the rate of isoproturon mineralisation was enhanced in most soil samples and the number of soil samples showing a low isoproturon mineralisation capacity decreased. A significant relationship (correlation coefficient, 0.89) between the logarithm of the mineralisation rate (ln k) and the soil pH was found, with a particularly strong positive effect on isoproturon mineralisation for pH > 6.5.