BACKGROUND: Stroke profoundly affects quality of life (QOL), including loss of employment, reduced social activity, shrinking social networks and low mood. Dysgraphia (impaired writing) is a common symptom of aphasia yet is rarely targeted in rehabilitation. Recent technological advances might challenge this, since much communication is now conducted digitally through writing. The rehabilitation of writing may therefore help to address the wider consequences of stroke and aphasia. AIMS: Can assistive technology (AT) training for people with dysgraphia: (1) improve written output, and are gains achieved only with AT? (2) improve reading comprehension scores, and are gains achieved only with AT? and (3) affect social participation, mood or QOL METHODS AND PROCEDURES: DESIGN: A mixed-methods, repeated measures, small group study design was adopted (qualitative outcomes will be reported elsewhere). PARTICIPANTS: Recruited from community settings, for example, Stroke Association communication support groups. INCLUSION CRITERIA: over 18 years old, aphasia due to stroke, acquired dysgraphia, writing more impaired than speech, fluent English prior to stroke, access to computer and Internet. EXCLUSION CRITERIA: currently receiving speech and language therapy, significant cognitive impairment, neuromuscular/motor-speech impairments/structural abnormalities, developmental dyslexia, uncorrected visual/auditory impairments. PROCEDURES: Screening and diagnostic assessments at time T1 (first baseline). Outcome measures at T1; repeated at T2 (second baseline), T3 (end of intervention), T4 (3-month follow up). Social participation assessment and cognitive monitoring at T2, T3, T4. INTERVENTION: Seven-ten hours individual therapy weekly and additional email support. Participants were trained to operate Dragon NaturallySpeaking (speech to text package) and ClaroRead (read writing aloud). Outcome measures were administered on pen and paper (control) and on computer, with AT enabled only at T3, T4. OUTCOMES AND RESULTS: Computer narrative writing was significantly improved by AT training (Friedman's χ (3) = 8.27, p = 0.041), indicating a compensatory effect of AT. Though reading comprehension significantly improved in the computer condition (Friedman's χ (3) = 21.07, p = 0.001), gains could not be attributed to the AT. Gains were achieved only when measures were administered on the keyboard, with AT enabled. Thus, a compensatory rather than remediatory effect was suggested. Social network size significantly increased; there were no significant changes in mood/QOL. Individual success rates varied. CONCLUSION AND IMPLICATIONS: The customisable AT training was acceptable to participants and resulted in significantly improved narrative writing. Compensatory AT interventions are a useful adjunct to remediatory writing interventions and may particularly support functional writing. WHAT THIS PAPER ADDS: What is already known on this subject Writing is rarely spared in aphasia and may present as the most impaired communication modality. Yet, people with aphasia report that writing is seldom included in their rehabilitation. Many communication activities are now conducted digitally through writing, therefore rehabilitation of this is more important than ever before. This study sought to address whether an assistive technology (AT) software package can improve writing and whether any changes were compensatory or remediatory. What this study adds to existing knowledge This group study found that AT training led to gains in written discourse and social network in people with aphasia and dysgraphia. Gains were not replicated in handwritten tasks, suggesting this was a compensatory therapeutic approach. What are the clinical implications of this work? AT programs such as this may present speech and language therapists with a practical, pragmatic adjunct to writing or typing therapy, particularly for clients with chronic, intractable impairments for whom remediatory therapy may have a low chance of success.