A meta-analysis of English proficiency and academic achievement in English-medium instruction

English-medium instruction (EMI) has expanded on the assumption that students’ English proficiency underpins their academic success, yet primary evidence is strikingly inconsistent. This study reports a random-effects correlational meta-analysis of the zero-order association between English proficiency and academic achievement across 29 studies ( N = 8396) in EMI higher education in six country/context groupings, with every effect size verified against its primary report and the evidence base current to 2026. The pooled correlation was moderate and positive, r = 0.34 (95% CI [0.26, 0.42]), and robust to publication bias, study quality, and dependencies among overlapping research teams. Heterogeneity, however, was substantial ( I 2 = 88.6%), and the 95% prediction interval (−0.07 to 0.65) indicates that the true association ranges from negligible to strong across settings. Academic discipline was the only significant moderator: the correlation was moderate in mixed-discipline samples ( r = 0.38) but weak in engineering samples ( r = 0.12), though this subgroup comprised only two studies, while neither proficiency-measure type nor outcome type moderated the effect. The findings position English proficiency as a modest, context-dependent correlate of EMI success that is positive on average, and they carry a direct message for English for Academic Purposes (EAP) practice. Because proficiency at entry accounts for only a modest, uneven share of the variance in achievement, admission thresholds cannot substitute for sustained, discipline-sensitive academic-language support during study. EAP provision calibrated to the language demands of specific disciplines is likely to serve EMI students better than uniform proficiency requirements.