The ability of profoundly deaf schoolchildren to learn the voicing distinction was studied over a period of 6 months. The children were given daily speech therapy and half of them received, in addition, visual feedback from a microprocessor-based speech training aid - the Fricative and Timing Aid. Their performance was analysed in terms of the acquisition of voiced v. voiceless stop consonants and of their position in a word. Results yielded by perceptual tests showed a substantial increase in the intelligibility of stop consonants at the end of the training programme. Retention scores, i.e. intelligibility scores obtained by re-testing the children after a 2-month period of non-practice, showed that practically no articulatory skills were forgotten during this interval. Measurements of vowel duration in word-final stop consonants indicated a considerable decrease in vowel length with training, and although this remained higher than for hearing children it probably contributed to higher intelligibility scores. Comparisons between children trained with or without visual feedback showed a definite advantage for the former. No improvement in intelligibility was found when no systematic speech training was available. It was concluded that while the voicing distinction may not be acquired spontaneously, it is a skill within reach of profoundly deaf children when they are trained adequately.