When rock phosphates (RP) are used to remediate Pb-contaminated soils, their effectiveness is likely affected by their grain size. In this study, the effect of grain size of rock phosphate on the effectiveness of heavy metal immobilization in two contaminated soils was measured in pot experiment. Rock phosphate was used with four different grain sizes: < 35, 35-72, 72-133 and 133-266 mu m. The application rate of rock phosphate in two soils was determined based on P/metals (Pb, Zn, Cu and Cd) molar ratio of 5.0 in the soils. The results showed that rock phosphate of the smallest grain size (< 35 mu m) was superior to all of other grain sizes more than 35 mu m for reducing uptake in plant (Brassica oleracea L.) shoots for Cd (19.6-50.0%), Pb (21.9-51.4%) and Zn (22.4-34.6%), respectively, as compared with the soil without application of rock phosphate. Sequential extraction analysis indicated that rock phosphate was most effective for soil Pb to induced transformation from non-residual fractions to a residual fraction than that for Zn and Cd. Such transformation was probably through dissolution of Pb associated with exchangeable (EX), organic fraction (OC), acidic fraction (AC) and amorphous Fe and Al oxides-bound (OX) fraction and precipitation of pyromorphite-like minerals. Results suggested that the rock phosphate with small grain size was superior to that with large grain size for in situ remediation technology. (c) 2005 Elsevier B.V. All rights reserved.