Cognitive remediation (CR) has been shown to improve cognitive abilities following a stroke. However, an updated quantitative literature review is needed to synthesize recent research and build understanding of factors that may optimize training parameters and treatment effects. Randomized controlled trials of CR were retrieved from seven electronic databases. Studies specific to adult stroke populations were included. Treatment effects were estimated using a random effects model, with immediate and longer-term follow-up outcomes, and moderator effects, examined for both overall and domain-specific functioning. Twenty-two studies were identified yielding 1098 patients (583 in CR groups). CR produced a small overall effect (g = 0.48, 95% CI 0.35-0.60, p < 0.01) compared with control conditions. This effect was moderated by recovery stage (p < 0.01), study quality (p = 0.04), and dose (p = 0.04), but not CR approach (p = 0.63). Significant small to medium (g = 0.25-0.75) post-intervention gains were evident within each individual outcome domain examined. A small overall effect (g = 0.27, 95% CI 0.04-0.51, p = 0.02) of CR persisted at follow-up (range 2-52 weeks). CR is effective and efficient at improving cognitive performance after stroke. The degree of efficacy varies across cognitive domains, and further high-quality research is required to enhance and sustain the immediate effects. Increased emphasis on early intervention approaches, brain-behavior relationships, and evaluation of activity and participation outcomes is also recommended.