Hair analysis, as complementary matrix, has expanded across the spectrum of toxicological investigations for misuse drug monitoring. Hair has become an important matrix for drug analysis, owing to the possibility to detect target analytes for long time periods, depending on hair length. A liquid chromatography-tandem mass spectrometry (LC-MS/MS) method has been developed for the quantitation of tramadol, a widely used centrally acting analgesic, and its main metabolites in hair (ODMT, NDMT, NOT). Hair samples were decontaminated and incubated overnight in diluted hydrochloric acid; the extracts were purified by mixed-mode solid phase cartridges and analyzed by LC-MS/MS in positive ionization mode monitoring two transitions per analyte. The procedure was fully validated in terms of linearity, limit of detection and lower limit of quantitation (LLOQ), accuracy, precision, recovery, matrix effect and selectivity. The linear regression analysis was calibrated by deuterated internal standards; for all analytes, responses were linear over the range 0.04-40.00 ng/mg hair, with R(2) values of at least 0.995. The method offered satisfactory precision (RSD < 10%), accuracy (90-110%) and recovery (> 90%) values. The found LLOQ values for tramadol and metabolites were in the range 0.010-0.030 ng/mg hair. The proposed procedure was successfully applied to quantify tramadol and metabolites in real hair samples submitted to our laboratory: three cases of tramadol assumption within the therapeutic dosage (3 × 2 segments) and one case of tramadol abuse in a binge pattern (8 segments). The ranges found for TRAM, ODMT, NDMT and NOT were markedly higher in the abuse case (63.42-107.30, 3.76-6.26, 24.88-45.66, 0.22-1.18 ng/mg hair, respectively) compared to the other case reports (3.29-20.12, 0.28-1.87, 0.45-4.32, 0.07-0.80 ng/mg, respectively); also the values of NMDT/ODMT ratio differed significantly. According to the obtained data, we hypothesized that the binge pattern may influence the metabolites' to parent drug concentration ratios; therefore this parameter could represent a target assessment tool to monitor abuse cases.