Two key arsenic metabolites, monomethylarsonous acid (MMA^III) and dimethylarsinous acid (DMA^III), have recently been detected in human urine. There is an increasing interest in the speciation of these metabolites in humans because of their demonstrated effects on cellular toxicity and DNA damage. However, there is no information on the oxidative stability of these arsenic species. It is not known whether and to what extent these trivalent metabolites are changed during sample handling and storage. The objective of this study was to demonstrate the oxidative conversion of these arsenic species during sample storage. We compared the effects of the storage temperature (25, 4, and −20°C) and storage duration (up to 5 months) on the stability of MMA^III and DMA^III in de-ionized water and in human urine. We used HPLC with hydride generation atomic fluorescence detection for the speciation of arsenic. This method provided sub-µg L⁻¹ to low-µg L⁻¹ detection limits for each arsenic species. We found that the oxidation of MMA^III and DMA^III was matrix and temperature dependent. Low temperature conditions (4 and −20°C ) improved the stability of these arsenic species over the room temperature storage condition. MMA^III in de-ionized water was relatively stable for almost 4 months, when stored at 4 or −20°C with less than 10% of MMA^III oxidized to MMA^V. In contrast, most of MMA^III (>90%) in urine was oxidized to MMA^V over the 5 month period under the 4 or −20°C storage condition. At 25°C, MMA^III in urine was completely oxidized to MMA^V within a week. DMA^III in de-ionized water was stable for only 2–3 days, being rapidly oxidized to DMA^V. DMA^III in urine was completely oxidized to DMA^V within a day at 4 or −20°C. The conversion of DMA^III to DMA^V in urine at 25°C was complete in 17 h. These results show that MMA^III and DMA^III are much less stable than other arsenic species, and their stability depends on sample matrix and temperature.