OBJECTIVE: The evidence for the effectiveness of memory rehabilitation is inconclusive. The aim was to compare the effectiveness of two group memory rehabilitation programmes with a self-help group control. DESIGN: Single-blind randomized controlled trial. PARTICIPANTS: Participants with memory problems following traumatic brain injury, stroke or multiple sclerosis were recruited from community settings. INTERVENTIONS: Participants were randomly allocated, in cohorts of four, to compensation or restitution group treatment programmes or a self-help group control. All programmes were manual-based and comprised two individual and ten weekly group sessions. MAIN MEASURES: Memory functions, mood, and activities of daily living were assessed at baseline and five and seven months after randomization. RESULTS: There were 72 participants (mean age 47.7, SD 10.2 years; 32 men). There was no significant effect of treatment on the Everyday Memory Questionnaire (P = 0.97). At seven months the mean scores were comparable (restitution 36.6, compensation 41.0, self-help 44.1). However, there was a significant difference between groups on the Internal Memory Aids Questionnaire (P = 0.002). The compensation and restitution groups each used significantly more internal memory aids than the self-help group (P < 0.01). There were no statistically significant differences between the groups on measures of mood, adjustment and activities of daily living (P > 0.05). CONCLUSIONS: There results show few statistically significant effects of either compensation or restitution memory group treatment as compared with a self-help group control. Further randomized trials of memory rehabilitation are needed.