A comparison of dynamic gas purging headspace-mass spectrometry and total vaporisation-mass spectrometry for the on-line analysis of liquid process streams was made. The detection limits (3s) for acetone in an aqueous matrix for the dynamic headspace and total vaporisation techniques are 0.4 and 11.2 ml ml 21 , respectively. The relative standard deviations for the determination of 100 ml ml 21 and 10% aqueous acetone solutions, respectively, are 9.5 and 1.9% for the dynamic headspace technique and 4.4 and 3.7% for the total vaporisation technique. Analysis frequencies of 20 and 60 h 21 were achieved for the dynamic headspace and total vaporisation techniques, respectively. An accurate analysis of acetone present in an 11 component process sample was achieved using the total vaporisation technique. The complexity of the sample matrix did not allow an accurate analysis to be performed using the dynamic headspace technique.