Summarizing the obtained results, the heptafluorobutyric derivatives of MOR and COD were analyzed by MS/MS-SRM, and HFBA 6-MAM by MS-SIM. Detection limits were evaluated by analyzing samples at known analyte amounts, at decreasing concentration: the concentration giving rise to an average value of S/N ratio of 3, and with a calculated concentration within 30% of the target value, was considered as the minimal concentration detectable. The obtained detection limits were: HFBA MOR, 3 pg/μl; HFBA COD, 5 pg/μl (S/N ± SD = 2.6 ± 0.8); HFBA 6-MAM, 2 pg/μl (S/N ± SD = 3.0 ± 0.4). In the case of MOR, both the SIM and the SRM techniques give the same detection limits, but they are referred to the analysis of standard solutions. If the method were applied to samples originating from complex matrices, tandem mass spectrometry would be preferable because it results in a reduction in the background noise (generally higher in biological matrices) and assures more specificity to the analysis. The experiments described above were performed in negative chemical ionization mode, but heptafluorobutyric derivatives were also analyzed by electron ionization (EI). As expected, detection limits were better than those observed in EI, because the use of negative ionization generally involves a decrement in background noise and, as a consequence, an increment in the S/N ratio. Actually, HFBA was chosen as derivatizing agent because it implies the addition of 7 or 14 fluorine atoms, which make possible the detection of analytes by negative chemical ionization. Since drugs accumulate in keratins and forensic toxicology has recently shown considerable interest in the determination of opiates in hair, hair was considered as biological matrix and analytes were extracted from matrix after acid digestion and solid-phase extraction (SPE) procedures, already reported in the literature and slightly modified in this study. Briefly, each hair sample (50 mg aliquots) was added to 0.1 N HCl, digested (45 °C, 18 h), and the solution was neutralized (0.1 M phosphate buffer, pH 6.5). After centrifugation (3500 rpm for 5 min), the supernatant was applied onto SPE columns (Bond Elute Certify, 130 mg; Varian, Harbor City, CA, USA), previously conditioned with methanol and phosphate buffer. After the washing steps (with deionized water, 0.1 N HCl and methanol) analytes were eluted by adding a freshly prepared solution of dichloromethane/isopropanol/ammonia (78:20:2, v/v/v), at a flow rate of 1 ml/min. Samples were dried under a slight flow of nitrogen, derivatized and analyzed as previously described. This analytical procedure was applied for the construction of a calibration curve, in order to verify linearity and to quantify real samples. The construction of the calibration curve requires the preparation of a series of standard samples: 50 mg aliquots from 'blank hair' were spiked with 50 μl of methanolic solutions of analytes at known concentration (0.156, 0.312, 0.625, 1.25, 2.50 and 5.00 ng/μl) and with nalorphine used as internal standard (5 ng/μl, methanolic solutions). Then samples were treated as described and analyzed by GC/NICI-MS. Each sample was analyzed in triplicate and data were elaborated using Xcalibur™ software (ThermoFinnigan). The concentration range used -(0.156-5.00 ng of analyte per mg of hair) was based on the amount of analytes expected from literature data, and on the 'cut-off' levels, i.e. the lower concentration levels below which a sample is considered negative, even if trace levels of analytes are actually found (hair cut-off level: 0.2 ng/mg). The following equation curves were obtained: MOR, y = 0.015 x + 0.0108 (correlation coefficient = R2 = 0.983); COD, y = 0.027 x + 0.0014 (R2 = 0.993); and 6-MAM, y = 0.0756 x + 0.0248 (R2 = 0.982). The values of correlation coefficients obtained for the three analytes under investigation showed good response linearity.