Different chemical modifiers for use with electrothermal atomic absorption spectrometry (ET AAS) were investigated in relation to determining the selenium in human urine samples. The samples were diluted in a solution containing 1% v/v HNO3 and 0.02% m/v cetyltrimethylammonium chloride (CTAC). Studying the modifiers showed that the use of either Ru or Ir as the permanent modifier gave low sensitivity to Se and the peak shape was very noisy, while Zr or Rh gave no peak at all. The same occurred when Zr was used in solution. For mixtures of permanent modifiers, Ir plus Rh or Zr plus Rh gave very low sensitivity, Zr plus Rh with co-injection of Ir in solution was also not efficient, Zr plus Rh in solution gave good sensitivity, but the best results were obtained with a mixture of Zr and Rh as the permanent modifier and co-injection of Rh in solution. Using this last modifier, the following dilutions with the HNO3 and CTAC were studied: 1:1, 1:2, 1:3 and 1:4. The best dilution was 1:1, which promoted good sensitivity and a more defined peak shape and made it possible to correct for the background using a deuterium arc lamp. Under these conditions, a characteristic mass of 26+/-0.2 pg was obtained for Se in aqueous solution. Six certified urine samples were analyzed using matrix matching calibration and the measured concentrations were in agreement with the certified values, according to a t-test at the 95% confidence level. Recovery tests were carried out and the recoveries were in the range 100-103%, with relative standard deviation better than 9%. The limit of detection (LOD, 3 sd, n = 10) was 3.0 microg L(-1) in the sample. The treated graphite tube could be used for at least 600 atomization cycles without significant alteration of the analytical signal.