Abstract This article investigates certain aspects of regressive voicing assimilation by means of a quantitative acoustic study of British English obstruent clusters. It is found that the phonologically voiceless obstruents /t, s/, and contrary to many impressionistic descriptions, the phonologically voiced sounds /z/ and to some extent, /d/ trigger certain forms of voicing assimilation in preceding obstruents. The precise patterning of the phonetic effects of the processes involved suggests that they should be modelled in terms of the coarticulation of the gestures that underpin the acoustic correlates of phonological voicing contrast.