Abstract <p>The results of microextraction separation and ICP–MS determination of inorganic arsenic species in natural waters are presented. The necessity of the separate quantification of analytes is justified, as arsenites exhibit toxicity dozens of times higher than arsenates. Separation was performed by the selective liquid–liquid microextraction of As(III) complexes with sodium diethyldithiocarbamate into an organic phase. Extraction conditions were optimized to achieve the highest recovery of As(III) complexes at approximately 95%. The As(III) complexes with sodium diethyldithiocarbamate were extracted into the organic phase using carbon tetrachloride as an extractant and methanol as a dispersant. Matrix effects of elements on analyte extraction from water were eliminated by performing a double microextraction of the analytes. Total inorganic arsenic and As(V) concentrations were determined by an ICP–MS analysis of the original water samples and their aqueous extracts obtained after the separation of inorganic arsenic species. The concentration of As(III) in water was calculated as the difference between total arsenic and As(V) concentrations. The limits of detection for As(III) and As(V) in water were equal, at 0.010 μg/L, within a linearity range of 0.05 to 100 μg/L, <i>R</i>² = 0.9998. The accuracy of the determination of inorganic arsenic species in water was confirmed by the standard addition method. The analytical procedure was validated using model waters and real samples of drinking and natural waters.</p>