Abstract Laboratory and field experiments were carried out on a loam soil and a sandy loam soil to study the persistence of metolachlor, atrazine and deethylatrazine in soil under Spanish conditions of maize cultivation. Herbicide levels in soil were studied for about 90 days at two different soil depths: 0–10 cm and 10–20 cm. Field half‐lives varied from 23 to 57 days for atrazine and from 22 to 42 days for metolachlor in the first layer. The highest deethylatrazine level in the first layer was 8% of the initial atrazine concentration. Maximum concentration in the second layer ranged from 5% to 17% for atrazine, from 3.5% to 31% for metolachlor and from 0.75% to 4.5% for deethylatrazine, in the conditions studied. Results were compared with outputs of two versions of the simulation model LEACHMV3: LEACHP and LEACHA. Laboratory assays were performed at different conditions to determine the atrazine and metolachlor degradation rates and the formation and disappearance rate of deethylatrazine, which were used in the computer programs. LEACHP predicted the herbicide persistence in soil with reasonable accuracy in the studied fields. Values from LEACHA model were also in agreement with the experimental data for these herbicides in one field, but in the other field, under a higher irrigation, this model overestimated the residues of these herbicides in the first layer.