BACKGROUND: Increasingly, computerised communication aids are used by people with severe, chronic aphasia. Although the candidacy for these devices is relatively unknown, it has been hypothesised that cognitive deficits have a negative impact on the functional use of Augmentative and Alternative Communication (AAC). Deficits of executive functioning are assumed to be particularly important, but other functions, such as memory and semantic processing, may also be relevant. In a previous study (van de Sandt-Koenderman, Wiegers, Wielaert, Duivenvoorden and Ribbers, in press) we reported the functional effect of TouchSpeak (TS), a computerised communication aid, in a group of stroke patients with severe aphasia. The successful participants showed different levels of proficiency. Some were able to use the system independently and creatively in many situations, some used it independently for trained situations and others remained partner dependent in using TS. AIMS: To find factors associated with the functional success of TS in people with severe aphasia, focusing on memory, executive functioning, semantic processing and communication skills. METHODS AND PROCEDURES: The data of 30 patients with severe aphasia were analysed retrospectively. All were trained to use TS in two self-chosen communicative situations. Four outcome levels were differentiated: no use, dependent use, independent use and extensive use of TS. Pre-training assessment included memory, executive functioning, semantic processing, and communication skills. The four outcome groups were compared regarding age, time post onset, gender, and aphasia type. The role of the cognitive variables was analysed with univariate ANCOVAs with contrast analysis, with correction for age, gender, aphasia type, or time post onset in case of significant differences between the groups on these variables. OUTCOMES AND RESULTS: Seven participants were classified as extensive users of TS, five were independent TS users and five were partner dependent. In 13 cases there was no functional use of TS. Extensive users were younger than the other outcome groups. Independent of this age effect, there was an effect of semantic processing; the no-use group scored significantly lower on semantics than all other groups. CONCLUSIONS: Only a minority of patients with severe aphasia may be expected to become independent, flexible users of high-tech AAC. The finding that functional success was related to semantic processing is clinically important. Prospective studies are needed to support the predictive value of semantic processing for high-tech AAC use. The importance of intact executive functioning is not supported in this study. The broad concept of "executive functioning" needs to be studied in more detail in relation to aphasia.