This study evaluated remediation of the herbicide sulfentrazone in soils with three different mineralogies (kaolinite, hematite, and gibbsite) and three remediation sulfentrazone treatments (Canavalia ensiformis L., Crotalaria juncea L., and natural attenuation). This study was conducted in a factorial scheme, in triplicate with randomized block design. Sulfentrazone was applied at 0 and 400 g ha⁻¹. We analyzed sulfentrazone residue in the soils by high-performance liquid chromatography and confirmed the results with bioassays of Pennisetum glaucum. Herbicide movement was greater in the kaolinitic soil without plant species. The retention of herbicide in the kaolinitic soil occurred in larger quantities in the 0–12 cm layer, with higher levels found in the treatments with plants. In the hematitic soil with C. juncea, all applied herbicides were concentrated in the 0–12 cm layer. In the other hematitic soil treatments, sulfentrazone was not detected by chemical analysis at any soil depth, although in many treatments, it was detected in the bioassay. Phytoremediation was more efficient with C. ensiformis grown in gibbsitic soil, reducing the sulfentrazone load by approximately 27%. Natural attenuation was more efficient than phytoremediation in oxidic soils due to soil pH and texture soils favored microbial degradation of the compound.