Tobacco dependence reaches one-third of the world population, and is the second leading cause of death around the world. Cotinine, a major metabolite of nicotine, is the most appropriate parameter to evaluate tobacco exposure and smoking status due to its higher stability and half-life when compared to nicotine. The procedure involves liquid-liquid extraction, separation on a RP column (Zorbax XDB C(8)), isocratic pump (0.5 mL/min of water-methanol-sodium acetate (0.1 M)-ACN (50:15:25:10, v/v/v/v), 1.0 mL of citric acid (0.034 M) and 5.0 mL of triethylamine for each liter) and HPLC-UV detection (261 nm). The analytical procedure proved to be sensitive, selective, precise, accurate and linear (r>0.99) in the range of 5-500.0 ng/mL for cotinine. 2-Phenylimidazole was used as the internal standard. The LOD was 0.18 ng/mL and the LOQ was 5.0 ng/mL. All samples from smoking volunteers were collected simultaneously to establish a comparison between serum, plasma, and urine. The urinary cotinine levels were normalized by the creatinine and urine density. A significant correlation was found (p<0.01) between all matrices. Results indicate that the urine normalization by creatinine or density is unnecessary. This method is considered reliable for determining cotinine in serum and plasma of smokers and in environmental tobacco smoke exposure.