This study was designed to test the explanation of the effects of unattended speech on serial recall put forward by Salamé & Baddeley (1982). As in their study, the effects of unattended speech were examined under conditions of articulatory suppression but, whereas Salamé & Baddeley used visual presentation of target items, the study reported here employed auditory presentation. In Expt 1, which used a between-subjects design and a nine-digit list, there were no effects of unattended speech when articulation was suppressed. Experiments 2 and 3 employed a within-subjects design and attempted to make the experimental task rather less exacting. Under these conditions, an unattended speech effect was observed, regardless of whether articulatory suppression occurred during presentation only or whether it continued during recall. It is argued that, consistent with the views of Salamé & Baddeley, unattended speech produces interference to a phonological input store rather than to the process of articulation itself.