Although the role of vanadium in the body has not been yet well defined, it is known that it inhibits the ATPase activity in erythrocytes and renal tubular cells, affecting the diuretic function of the kidneys, the metabolism of leukocytes and macrophages, the function of the heart muscle and causes insulin-like effects in pharmacological doses. Vanadium is of great importance to the industry, and is mostly used as catalyst, and in the production of glass and steel. Workers of the steel industries as well as those in plants that use large amounts of fossil fuels are potentially exposed to large amounts of vanadium, and it is important to monitor their serum levels. In the present study it was developed a simple and direct method for determination of vanadium in serum using graphite furnace atomic absortion spectrometry optimized by a central composite design. External calibration was possible, using calibration solutions prepared in 2 % v/v HNO3. The sample was 2+1 diluted in the same medium and the calibration curve was found linear up to at least 3 ug L(-1). The limits of detection and quantification were respectively 0.3 and 1.0 ug L(-1), in the original sample. Average recoveries were 97.2%, 99.2% and 105.3% for concentrations of 1, 2 and 3 ug L(-1), respectively. Excellent agreement was found between found and certified values in the analysis of a certified reference material. The sample throughput was 8 h-1 (n=3). In the study of repeatability, the largest coefficient of variation was 10.3% in the sample spiked with 1 ug L(-1) of vanadium. In the study of the intermediate precision the higher coefficient of variation was 5.6% in the sample spiked with 2 μg.L-1. The combined uncertainties obtained were 1.65x10-1, 2.72x10-1 and 3.72x10-1, for concentrations of 1, 2 and 3 ug L(-1), respectively. The graphite tube life time in the optimized conditions was 150 cycles, approximately. The method was considered adequate to the routine determination of vanadium in serum in clinical laboratories.