The ability of an adult with profound hearing impairment to integrate speech information from touch, aided hearing, and speechreading in identification of open‐set words was investigated. A list was obtained of 735 words that the subject failed to identify using any single modality: touch, with either the Tacticon 1600, a multichannel electrocutaneous vocoder (TV), or the Tactaid II, a 2‐channel vibrotactile aid (TA); aided hearing (H); or speechreading (S). To test integration, observed word identification performance in combined‐modality conditions was compared with predicted performance calculated from single‐modality scores. Words were randomly assigned to seven conditions: (a) S+H, (b) H+TV, (c) H+TA, (d) S+TV, (d) S+TA, (e) S+TV+H, and (f) S+TA+H. Results indicated that the subject integrated speech information across modalities, with highest performance in the S+TV+H and S+TA+H conditions. Integration also occurred when both speechreading and touch were used and when both speechreading and aided hearing were used.