This investigation was undertaken to identify the most significant parameters that can be used to model the bioaccessibility of lead in contaminated soils. Twenty-five soil samples contaminated by two distinct sources of contamination (mining and smelting areas) were investigated in this work. Edaphic soil properties were determined using the classic techniques. Speciation and solid phase distribution of Pb were determined by sequential fractionation and physical analysis (XRD and SEM-EDX). Bioaccessibility of Pb was measured using the unified BARGE protocol on each soil samples and each residue following the different steps of the sequential extraction procedure. Results showed that Pb speciation varied according to the source of contamination. Pb gastric bioaccessibility ranged from 12% to 86% in mining soils and from 40% to 95% in smelting soils. Soil edaphic properties (pH, CEC, clays content or organic matter content) cannot be used to model Pb bioaccessibility as these parameters could not be extrapolated from on site to another. The differences in Pb bioaccessibility between the 25 soils were better reflected by the nature of the Pb-bearing minerals as characterized by sequential extractions. Soluble, exchangeable and acid-soluble phases could explain from 70 to 98% of Pb bioaccessibility. Consequently, Pb bioaccessibility is extremely sensitive to the Pb-bearing phase distribution. This distribution should be thoroughly evaluated before applying any remedial techniques that would potentially affect Pb speciation in soils.