The arsenic (As) spectral interference observed in the determination of cadmium (Cd) by inductively coupled plasma atomic emission spectrometry (ICP-AES) was studied in atomic absorption spectrometry (AAS) using flame (FAAS) and graphite furnace (GFAAS) as atomizers. The soils of 15 kitchen gardens located near two smelters in the North of France were selected according the ratio As/Cd. Four different extracting solutions usually used to evaluate the mobility of Cd were chosen to extract Cd from these soil samples: citric acid 0.11 M, acetic acid 0.11 M, calcium chloride 0.01 M and water. The quantitative determinations of Cd in the 15 soils for each solvent were investigated by ICP-AES at two lines (228.802 and 214.438 nm) and by FAAS or GFAAS with two-way background compensation. Compared to the Cd concentrations measured in the acid solutions and in the CaCl(2) solution after the addition of a chemical modifier, it was clearly demonstrated that the high-speed self-reversal background compensation (HSSR-method) was the method of choice to eliminate the spectral interference of As during Cd determination by FAAS and by GFAAS. In water, it was shown that the deuterium lamp used for the background compensation (D(2)-method) was able to eliminate the most of the As interference. In comparison with Cd concentrations in water after adding a chemical modifier, those obtained with the HSSR-method were similar and a very good correlation was obtained between these two methods (R(2)=0.995). It was therefore established that the HSSR-method would be able to replace the chemical modifiers to eliminate As interference in the determination of Cd-extractable from As contaminated soils.