The performance of gas chromatography coupled to mass spectrometry in the positive chemical ionization mode using ammonia as reagent gas (GC-PCI-MS) in the analysis of C37 alkenones for paleotemperature estimation has been re-evaluated. In some conditions, the discrepancies observed in the measurement of the U37(k') index with this technique as compared with GC equipped with flame ionization detection (GC-FID) cannot be explained by differences in sensitivity between the tri- and diunsaturated alkenones. Thus, at low (currently <0.3) or high (currently >0.4) U37(k') values the GC-PCI-MS determinations may be observed to be lower or higher, respectively, than those measured with GC-FID. As shown by analysis of a series of synthetic C37 alkenone standards these discrepant results can be explained by non linear effects in the GC-PCI-MS response factors. Second-order polynomial functions provide equations that describe better the signal to amount of analyte ratios. Users of GC-PCI-MS should calibrate their instruments with standards of known C37 alkenone composition in order to minimize non-linear effects.