This study synthesised previous literature focused on college English for Academic Purposes (EAP) programmes for writing. A meta-analysis of identified longitudinal studies (14 samples; 175 effect sizes; N = 602) revealed that the students improved their English writing through EAP programmes (d = 0.45, SE = 0.07, p < 0.001). By computing effect size mean estimates from ranges for the 25th, 50th, and 75th percentiles, we proposed effect size interpretation benchmarks for this subdomain of 0.3, 0.5, and 0.7 for small, medium, and large effects, respectively, for interpreting the improvements in writing through college EAP programmes. The meta-regression revealed that the students with low English proficiency improved significantly more than those with higher proficiency. Further, the programmes had a small effect in improving accuracy and complexity dimensions of writing. Different features of programmes and contexts have also been found to influence the degree of the improvements in complex ways.