The objective of this study was to examine the effect of biosurfactant on sorption of phenanthrene (PHE) onto the original or H2O2-treated black loamy soil (typic isohumisols) and red sandy soil (typic ferralisols). The sorption isotherms were performed with the original and "soft" carbon-removed soils in the presence and absence of biosurfactant (200 mg/L). The sorption and degradation of biosurfactant were investigated. The result showed that organic matter played an important role in PHE sorption onto the black loamy and red sandy soils, and the PHE sorption isotherms on the "soft" carbon-removed soils exhibited more nonlinearity than those on the original soils. The values of partition coefficient (Kd) on the original black loamy soil with or without 200 mg/L biosurfactant were 181.6 and 494.5 mL/g, respectively. Correspondingly, in the red sandy soil, Kd was 246.4 and 212.8 mL/g in the presence or absence of biosurfactant, respectively. The changes of Kd suggested that biosurfactant inhibited PHE sorption onto the black loamy soil, but facilitated PHE sorption onto the red sandy soil. The nonlinearity of PHE sorption isotherm was decreased in the presence of biosurfactant. Site specific sorption might occur during PHE sorption onto both the original and the "soft" carbon-removed soils in the presence of biosurfactant. It was noted that biosurfactant could also be sorbed onto soils. The maximal sorption capacity of the red sandy soil for biosurfactant was (76.9 ± 0.007) μg/g, which was 1.31 times that of black loamy soil. Biosurfactant was degraded quickly in the two selected soils, and 92% of biosurfactant were mineralized throughout the incubation experiment for 7 d. It implied that biosurfactant should be added frequently when the remediation of polycyclic aromatic hydrocarbon (PAH)-contaminated soils was conducted through PAH desorption approach facilitated by biosurfactant.