Laboratory batch experiments were carried out to study adsorption of arsenate on eight ferrallitic soils different in property. The Langmuir one-surface equation and the Langmuir two-surface equation was used to fit the data of arsenate adsorption, separately. Adsorption capacities of the soils were derived by the better fitting equation. Effects of soil properties on arsenate adsorption were explored through simple linear regression analysis. Results show that ferrallitic soils were high in arsenic adsorption capacity, all displaying non-linear adsorption isotherms, which were better fitted by the Langmuir two-surface equation than by the one-surface equation. The correlation coefficients between the adsorption capacities predicted by the two-surface equation and measured values (0.989 to 0.998), were higher than the determination coefficients obtained through prediction by the one-surface equation (0.935 to 0.978). The high adsorption capacity of the soils could be explained with the existence of high-energy and low-energy surface adsorption sites in the soils. Among all the soils tested, the red soil developed from sand shales in Kunming, Yunnan Province showed the highest arsenic adsorption capacity, which was 3 498 mg kg^-1, according to the calculation with the Langmuir two-surface equation. Contents of citrate-dithionite extractable Fe, total Al, total Fe, clay and total P exhibited more significant positive influence on arsenic adsorption capacity of the ferrallitic soils than any other soil properties.