Two measures of verbal inference-making ability (a sentence-completion lest; SCT, and a word-completion test; WCT) and one test of working memory capacity were examined in relation to speechreading performance. The results demonstrated that the SCT proved to be the only variable that was substantially correlated with speechreading performance. The contribution from working memory capacity and the WCT test to speechreading is mainly via their contribution to the SCT. Skilled SCT-performance was particularly tied to speechreading conditions with a low level of contextual information accompanying the speechreading task. No general or specific difference was found between hearing-impaired and normally-hearing subjects in the speechreading test, thus indicating that speechreading performance cannot be predicted by factors related to the hearing-impairment. Rather, information-processing skills seem to be decisive.