The mercury determination in gasoline by cold vapor and AAS, with reduction in organic liquids and in solution of three components (microemulsion)was studied. Some methodologies were used, with cold vapor generation and atomic absorption spectrometer, to know: direct determination in gasoline; direct determination in gasoline stabilized as microemulsion, with and without preconcentration in gold; and direct determination in gasoline stabilized as microemulsion using a trap for gasoline vapors before the preconcentration in gold. Analyzing the results, it was verified that the measurements from the gasoline stabilized as microemulsion supplies more repetitive readings than the gasoline in nature; moreover, it was observed that is necessary to use preconcentration in gold, due to the lower mercury levels in gasoline; however, it was detected that the gasoline vapors poison the gold. Amongst the studied traps to prevent the poisoning of the gold, the trap with solution of K2Cr2O7/ H2SO4 was chosen, in which, the vapors proceeding from the reaction bottle had been bubbled and the Hg0, oxidated; later, the solution was reduced with SnCl2/ H2SO4 and vapors sent to the gold. Thus, the results had indicated that the method for total mercury determination in gasoline, in which the mercury is reduced directly from the gasoline as microemulsion, using the K2Cr2O7/H2SO4 trap before the preconcentration in gold, with CV-AAS detection, is recommended. After multivariate optimization analytical curves showed coefficients of correlation as good as 0.9999 with a corresponding characteristic mass of 2ng of mercury. The limit of detection obtained for this method was 0.10 ng/mL (0.14 ng/g).