Metalaxyl was used to control Pythium diseases of carrots in experiments on farms with a history of cavity spot. The first experiment compared the method of application (sprayed, banded or broadcast) and rate (0, 1.5, 3 or 6 kg a.i. ha−1) one week after sowing. Three additional experiments compared the rate (0, 0.75, 1.5 or 3 kg a.i. ha−1) and time (sowing, 1‐ to 2‐ or 4‐ to 5‐true‐leaf stage) of application. In expt 1, the application of metalaxyl, but not the method by which it was applied, increased yield by 20% and significantly reduced the incidence of cavity spot, forking and misshapen carrots. In expts 2, 3 and 4, neither the rate nor time of application affected yield or reduced the incidence of Pythium diseases. Comparison of the sites showed that they differed in past metalaxyl usage. Metalaxyl had not been used on the site of expt 1, but had been used previously at sites 2, 3 and 4. Laboratory experiments were conducted to determine whether these differences in efficacy resulted from reduced sensitivity of Pythium isolates to metalaxyl, or reduced persistence of metalaxyl in soil. ED50 values showed that there was no reduction in metalaxyl sensitivity. The half‐life of metalaxyl was 82 days in soil from expt 1, but was 10 days or fewer in soils from expts 2, 3 and 4. Thus the failure of metalaxyl to control Pythium diseases was associated with reduced persistence in soil, not reduced sensitivity of the target fungi.