Leaching can be infl uenced by the physical characteristics of soil and precipitation when the herbicide S-metolachlor is used intensively in sugarcane areas. The objective of this study was to use bioindicator plants to evaluate the percolation potential of S-metolachlor when submitted to natural pluviometric precipitation in two soils of contrasting textures: clayish (Entisols) and medium sandy (Oxisols). Samples were collected using PVC columns that were introduced in both soil types after the herbicide had been applied at the doses of 1,960 and 1,440 g ha–1 a.i of S-metolachlor, recommended for the clayish and sandy soils, respectively. The columns were carefully extracted from the soil by excavating around them, thus keeping soil physical integrity. The samples were taken after the soils had been submitted to 30 – 40, 60 – 80, and 100 – 120 mm of rainfall. The columns were divided by a longitudinal cut to permit the sowing of seeds of the bioindicator plants, allowing suffi cient space for 25 plants per species in the columns (cucumber and lettuce). The bioindicator plants were evaluated with respect to their degree of toxicity by a degree scale ranging from 0 to 100%; these evaluations occurred at 5, 7, 10, and 12 days after sowing. We verifi ed that herbicide percolation depended on soil texture and accumulated rain. Under high precipitation conditions, the herbicide percolates to deeper depths in sandy soils. Cucumber was the most appropriate bioindicator plant in soils of clayish texture. In sandy soils, both cucumber and lettuce showed suitability as bioindicator plants.