Objectives: To estimate the efficacy of computer-based cognitive interventions for improving cognition in people with dementia (PWD). Method: Online literature databases were searched for relevant studies. Interventions were categorised as follows: cognitive recreation, cognitive rehabilitation, cognitive stimulation or cognitive training. A systematic review, quality assessment and meta-analyses were conducted. Results: Twelve studies were identified. Their methodological quality was acceptable according to Downs & Black criteria, the weakest methodological area being the external validity. The metaanalyses indicated cognitive interventions lead to beneficial effects on cognition in PWD (SMD ¡0.69; 95% CI D¡1.02 to ¡0.37; P < 0.0001; I2 D 29%), depression (SMD 0.47; 95% CI D 0.16 to 0.78; p D 0.003; I2 D 0%) and anxiety (SMD 0.55; 95% CI D 0.07 to 1.04; P < 0.03; I2 D 42%). They benefited significantly more from the computer-based cognitive interventions than from the non-computerbased interventions in cognition (SMD 0.48; 95% CI D 0.09 to 0.87; P D 0.02; I2 D 2%). Conclusion: Computer-based cognitive interventions have moderate effects in cognition, and anxiety and small effects in depression in PWD. No significant effects were found on activities of daily living. They led to superior results compared to non-computer-based interventions in cognition. Further research is needed on cognitive recreation and cognitive stimulation. There is also a need for longerterm follow-up to examine the potential retention of treatment effects, and for the design of specific outcome measures.