The effectiveness of isotopically and non-isotopically labelled internal standards in reducing matrix-induced effects is evaluated. The question is addressed in the quantitative analysis by gas chromatography/mass spectrometry (GC/MS) of dimethyl sulphide, chosen as a typical example of volatile sulphur compounds, in wine matrices. When matrix/run effects are not cancelled out the use of a variance component model (VCM) to handle the linear calibrations obtained by regression technique is successful. The method implies the estimation of an overall calibration straight line, which properly takes into account the uncertainty due to different matrices, the calibration run and the measurement error, making the use of an isotopically labelled internal standard not necessary. The obtained results show that the benefits of lowering times and costs for routine analyses compensate for the small increase in uncertainty in the concentration values obtained in the regression analysis and the slight increase in the detection limit.