A method for direct determination of manganese (Mn) in human serum by graphite furnace atomic absorption spectrometry (GFAAS) was proposed in this work. The samples were only diluted 1:4 with nitric acid 1% (v/v) and Triton(®) X-100 0.1% (v/v). The optimization of the instrumental conditions was made using multivariate approach. A factorial design (2(3)) was employed to investigate the tendency of the most intense absorbance signal. The pyrolysis and atomization temperatures and the use of modifier were available and only the parameter modifier use did not have a significant effect on the response. A Center Composed Design (CCD) presented best temperatures of 430 °C and 2568 °C for pyrolysis and atomization, respectively. The method allowed the determination of manganese with a curve varying from 0.7 to 3.3 μg/L. Recovery studies in three concentration levels (n=7 for each level) presented results from 98 ± 5 to 102 ± 7 %. The detection limit was 0.2 μg/L, the quantifying limit was 0.7 μg/L, and the characteristic mass, 1.3 ± 0.2 pg. Intra- and interassay studies showed coefficients of variation of 4.7-7.0% (n=21) and 6-8%(n=63), respectively. The method was applied for the determination of manganese in 53 samples obtaining concentrations from 3.9 to 13.7 μg/L.