Emissions from primary lead smelters have been recognized as one of the mainly factor which has contributed to the contamination of soils by metals. Less attention has been paid to volatile metalloids such as arsenic (As) which accompanies lead (Pb) smelting activities. One of the objectives of this study was to determine the As concentrations in various extracting solutions using a collection of urban soils located no far away from two former Pb and zinc plants in the North of France. The procedure for the determination of As, As(III), and As(V) with hydride vapor generator atomic absorption spectrometry was described in details. Pseudo-total concentrations of As in the studied soils ranged from 5.3 to 65.9 mg kg(-1). Good correlations were found between As and lead, zinc, and cadmium concentrations in soils. These depended on the soil uses and the soil distance from the source of contamination. Because the form of As may pose a health risk to human population, its speciation was determined in each urban top soils. Very good correlations were found between As(III) and As(V) versus As concentrations in soils studied, but the results did no permit to establish a relation between the location of soils and their uses. In contrast, it was shown that the highest mobility factor and lowest partitioning index values were related to the location. The mobilty of As depended on the assimilated phosphorus (P), carbonate contents, and pH. The percentages of the water-extractable As concentrations ranged from 0.3 to 3.0% of the As concentrations in soils. Very good positive correlations between water-extractable As(III) and As(V) versus water-extractable As concentrations were obtained. It was shown that the water-extractable As(III) concentrations depended on the soil uses. The results revealed that soils for which the As was the most mobile presented the highest water-extractable As concentrations. Principal component analysis indicated that mechanisms related to the release of As depended on the physico-chemical parameters of the soils, particularly on the assimilated P, organic matter, and/or iron oxides/hydroxides contents. Finally, the glasshouse experiments using ryegrass as plant model and three soils with similar physico-chemical parameters with regard to the PCA analysis showed that the water extracting solution could be a good indicator to evaluate the As phytoavailability.