Abstract.A comparative study of different chemical modifiers in graphite furnace atomic absorption spectrometry for the direct determination of mercury in sea water samples, in synthetic sea water sample of high (72.8%0) and low 34.2%0) salinity and in aqueous solutions, was carried out. The use of reduced palladium produces better results. The mixture of palladium nitrate and ascorbic acid, gives the best limit of detection (1.9 btgl-1). The use of reduced palladium and magnesium nitrate produced excellent recoveries (close to 100%) in the whole salinity range for all mercury concentration tested. The use of palladium nitrate alone or combined with magnesium nitrate gave good recoveries with respect to a real sea water sample for low salinities. The interference from the major components of sea water were completely removed by using reduced palladium and magnesium nitrate modifiers. Thus, a single calibration curve with synthetic sea water may be applied to the analysis of sea water samples of widely differing salinities.Key words: electrothermal atomization atomic absorption spectrometry, mercury, sea water, modifiers, palladium nitrate, magnesium nitrate, reduced palladium.Atomic absorption spectrometry with electrothermal atomization (ET-AAS) was not recommended for the determination of mercury owing to the possible losses of mercury during the drying and ashing steps, and the cold vapour is known as the more useful technique for mercury determination. However, the problem related to the loss of mercury in ET-AAS can be minimized with the use of chemical modifiers, which stabilize mercury at higher temperatures.