BACKGROUND: People with traumatic brain injuries (TBIs) commonly report memory impairments. These are persistent, debilitating and reduce quality of life, but patients do not routinely receive memory rehabilitation after discharge from hospital. OBJECTIVE: To assess the clinical effectiveness and cost-effectiveness of a group memory rehabilitation programme for people with TBI. DESIGN: Multicentre, pragmatic, cluster randomised controlled trial. Qualitative and health economic evaluations were also undertaken. SETTING: Community settings in nine sites in England. PARTICIPANTS: Participants were aged 18-69 years, had undergone a TBI > 3 months prior to recruitment, reported memory problems, were able to travel to a site to attend group sessions, could communicate in English and gave informed consent. RANDOMISATION AND BLINDING: Clusters of four to six participants were randomised to the memory rehabilitation arm or the usual-care arm on a 1 : 1 ratio. Randomisation was based on a computer-generated pseudo-random code using random permuted blocks of randomly varying size, stratified by study site. Participants and therapists were aware of the treatment allocation whereas outcome assessors were blinded. INTERVENTIONS: In the memory rehabilitation arm 10 weekly sessions of a manualised memory rehabilitation programme were provided in addition to usual care. Participants were taught restitution strategies to retrain impaired memory functions and compensation strategies to enable them to cope with memory problems. The usual-care arm received usual care only. MAIN OUTCOME MEASURES: Outcomes were assessed at 6 and 12 months after randomisation. Primary outcome: patient-completed Everyday Memory Questionnaire - patient version (EMQ-p) at 6 months' follow-up. Secondary outcomes: Rivermead Behavioural Memory Test - third edition (RBMT-3), General Health Questionnaire 30-item version, European Brain Injury Questionnaire, Everyday Memory Questionnaire - relative version and individual goal attainment. Costs (based on a UK NHS and Personal Social Services perspective) were collected using a service use questionnaire, with the EuroQol-5 Dimensions, five-level version, used to derive quality-adjusted life-years (QALYs). A Markov model was developed to explore cost-effectiveness at 5 and 10 years, with a 3.5% discount applied. RESULTS: We randomised 328 participants (memory rehabilitation,  = 171; usual care,  = 157), with 129 in the memory rehabilitation arm and 122 in the usual-care arm included in the primary analysis. We found no clinically important difference on the EMQ-p between the two arms at 6 months' follow-up (adjusted difference in mean scores -2.1, 95% confidence interval -6.7 to 2.5;  = 0.37). For secondary outcomes, differences favouring the memory rehabilitation arm were observed at 6 months' follow-up for the RBMT-3 and goal attainment, but remained only for goal attainment at 12 months' follow-up. There were no differences between arms in mood or quality of life. The qualitative results suggested positive experiences of participating in the trial and of attending the groups. Participants reported that memory rehabilitation was not routinely accessible in usual care. The primary health economics outcome at 12 months found memory rehabilitation to be £26.89 cheaper than usual care but less effective, with an incremental QALY loss of 0.007. Differences in costs and effects were not statistically significant and non-parametric bootstrapping demonstrated considerable uncertainty in these findings. No safety concerns were raised and no deaths were reported. LIMITATIONS: As a pragmatic trial, we had broad inclusion criteria and, therefore, there was considerable heterogeneity within the sample. The study was not powered to perform further subgroup analyses. Participants and therapists could not be blinded to treatment allocation. CONCLUSIONS: The group memory rehabilitation delivered in this trial is very unlikely to lead to clinical benefits or to be a cost-effective treatment for people with TBI in the community. Future studies should examine the selection of participants who may benefit most from memory rehabilitation. TRIAL REGISTRATION: Current Controlled Trials ISRCTN65792154. FUNDING: This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in ; Vol. 23, No. 16. See the NIHR Journals Library website for further project information.