In this study, plant (SAP, TAN) and microbial (RAM) biosurfactants were applied to two soils, in single or sequential soil washing. Soil 1 was moderately acidic, with higher content of organic matter and HMs (Cd, Cu, Ni, Pb and Zn), and was aged for 24 months. Soil 2 was alkaline, with lower content of organic matter and HMs, and was aged for 1 month. The stability of the HMs (as IR), except for Pb, was higher in soil 1 than in soil 2, but the rankings of HM stability were similar: Cu > Ni ≈ Pb > Zn > Cd (soil 1) and Cu > Pb > Ni > Zn > Cd (soil 2). The HMs were removed more efficiently from soil 2 than from soil 1, mainly from the readily available fraction and partially from the reducible fraction. Cu (0.3–0.5 times), Ni (0.4–1.1 times) and Zn (0.5–1.0 times) showed the largest increases in stability after soil washing with SAP. With a single washing, SAP and TAN were the least effective for Pb removal (18–31% and 11–35%, respectively), while RAM was the most effective for Cd (69–73%) and Pb (58–63%) removal. Sequential soil washing, especially with SAP followed by RAM, was the best choice for simultaneous removal of multi-HMs.