A method for the selective quantitative determination of inorganic arsenic [As(III) + As(V)] in seafood was developed. In order to do so, various procedures for the solubilization and extraction of inorganic arsenic quoted in the literature were tested. None provided satisfactory recoveries for As(III) and As(V) in real samples. Consequently, a methodology was developed which included solubilization with HCl and subsequent extraction with chloroform. The arsenic was solubilized in 9 mol l^–1 hydrochloric acid. After reduction by hydrobromic acid and hydrazine sulfate, the inorganic arsenic was extracted into chloroform, back-extracted into 1 mol l^–1 HCl, dry-ashed, and quantified by hydride generation-atomic absorption spectrometry (HG-AAS). The analytical features of the method are as follows: detection limit, 3.07 ng g^–1 As (fresh mass); precision (RSD), 4.0%; recovery, As(III) 99%, As(V) 96%. In the optimized conditions, other arsenic species—dimethylarsinic acid (DMA), arsenobetaine (AB), arsenocholine (AC) and tetramethylarsonium-ion (TMA⁺)—were not co-extracted. However, different percentages of minor species were extracted with chloroform: monomethylarsonic acid (MMA) 100%, and trimethylarsine oxide (TMAO) 3–10%. Real samples and reference materials of seafood (DORM-1, DORM-2, TORT-2, CRM-278 and SRM-1566a) were analyzed. The analysis of DORM-1 provided an inorganic arsenic value of 124 ± 4 ng g^–1 As, dry mass (dm), which is very close to the value obtained by other authors using high performance liquid chromatography-inductively coupled plasma-mass spectrometry (HPLC-ICP-MS) and ionic chromatography-hydride generation-atomic absorption spectrometry (IC-HG-AAS).