Rationale <p>Although there is broad agreement that caffeine provides an acute improvement in attention in the normal population, estimates of effect size vary and the relationship between dose and effect is unclear.</p> Objective <p>To examine the acute effect of caffeine on attention in a systematic review and meta-analysis.</p> Methods <p>PsycINFO, PubMed, and Scopus were searched for records published in English with no limits on the&#xa0;year of publication. Studies were included if they were randomized, double-blinded, placebo-controlled, and if they examined the acute effect of pure caffeine on behavioral tests of attention in rested, healthy adults. For every included trial, eligible outcomes were extracted and aggregated to form one composite standardized mean difference (SMD; Hedges’ adjusted g) for reaction time and one for accuracy. The SMDs were then combined in random-effects meta-analyses. Additionally, several subgroup analyses were conducted, including meta-regressions on dose-response relationships.</p> Results <p>Thirty-one trials with a total of 1455 participants were included in the meta-analysis. Significant effects in favor of caffeine were found for both accuracy, <i>g</i> = 0.27, and reaction time, <i>g</i> = 0.28. Subgroup analyses showed that higher doses of caffeine (≥ 200&#xa0;mg) improved both reaction time and accuracy more than lower doses, but whereas a positive linear dose-response relationship was found for reaction time, a quadratic relationship was found for accuracy. The effect of caffeine was not related to differences in habitual caffeine consumption, task complexity, or which attention network was taxed.</p> Conclusion <p>The current evidence shows that in the normal population, caffeine acutely enhances attention by improving both reaction time and accuracy. However, whereas higher doses continue to enhance reaction time, accuracy improves only up to a certain point before declining.</p>