Abstract Gas Chromatographic analyses, using ~105°C heated, on-line crushing as the release system, and photoionization (PID) and micro-thermal conductivity (TCD) detectors, of 17 samples (~2 g) in duplicate from three zones of the Tanco granitic pegmatite, SE Manitoba, indicate that H 2O and CO 2 generally account for 99 mol% of the volatiles detected from fluid inclusions in bulk. The balance of the volatiles comprises trace amounts of N 2, Ar, CO, CH 4, C 2H 4, C 2H 6, C 2H 2, COS, C 3H 6, C 3H 8, C 4H 8 (isobutylene), and n-C 4H 10, in order of retention time. Samples of vug quartz, beryl fringe, and some samples from the quartz zone show little variation in CO 2:CH 4 ratio from crush to crush, suggesting that the same fluids are released by both crushes. In general, the analyses of sample and duplicate are consistent. The volatile chemistries of all the samples are consistent with the results of fluid inclusion microthermometric and pétrographic observations (e.g., higher CH 4 in beryl fringe samples with lower CO 2 melting points of −57.3 ± 1.0°C; n = 61). The fluid trapped by the inclusions in a sample of lower intermediate zone (LIZ) vug quartz (PV-18-6-core) shows no petrographic or microthermometric evidence for H 2O-CO 2 phase separation and represents an internal, probably magmatically derived fluid. The volatile composition of this fluid is 95.6 mol% H 2O, 4.3 mol% CO 2, 436 ppm CH 4, 277 ppm N 2, 19 ppm Ar, 3 ppm C 2H 6, 1 ppm COS, 1 ppm C 3H 8, 0.3 ppm C 2H 4, 98 ppb C 2H 2, plus SO 2 and trace C 3H 6. This composition (i.e., H 2O-CO 2-CH 4-N 2; S species not included) is comparable to those of high-temperature (> 400°C) volcanic gases (e.g., White Island, New Zealand; Mt. St. Helens, USA) and is therefore compatible with a magmatic origin. The salinity of this fluid ranges from 0.31 to 1.96 mCl −1 ( x = 6.6 ± 1.3 (1 σ) eq. wt% NaCl) , which is higher than that predicted by the partition experiments of Webster and Holloway (1988) suggesting that the Cl solubility limit of the late stage Tanco melt may have been reached. H 2O-CO 2 phase separation has influenced the bulk chemistry of the volatiles released from the fluid inclusions in the vug quartz, beryl fringe, and quartz zone. Preferential trapping is indicated in the latter two cases by the presence of higher amounts of CO 2 than are present in the fluid prior to phase separation (LIZ vug quartz). Clear trends in the ratios of CO 2/ trace gases indicate that the trace volatiles partitioned during phase separation, resulting in N 2, CH 4, and the C 2+ hydrocarbons being preferentially incorporated into the CO 2-rich phase. The occurrence of such depletion trends implies that the pegmatite was partially open to CO 2 loss. The gases partition into the CO 2-rich phase in the following order: N 2 < CH 4 < C 3 H 8 < C 2 H 6 < C 2 H 4 < C 2 H 2 < C 3 H 6. CO 2/ volatile vs. CO 2 CH 4 slopes vary from as little as 0.7 ( CO 2 N 2 ) to as much as ~10 5 ( CO 2 C 3H 6 ); however, the behavior of the individual hydrocarbons is complex, reflecting the opposing or additive effects of cooling and phase separation.