The present meta‐analysis examined the overall average correlation (weighted for sample size and corrected for measurement error) between passage‐level second language (L2) reading comprehension and 10 key reading component variables investigated in the research domain. Four high‐evidence correlates (with 18 or more accumulated effect sizes: L2 decoding, L2 vocabulary knowledge, L2 grammar knowledge, first language [L1] reading comprehension), and six low‐evidence correlates (L2 phonological awareness, L2 orthographic knowledge, L2 morphological knowledge, L2 listening comprehension, working memory, metacognition) were included in the study. For the four high‐evidence correlates, a series of moderator analyses were also carried out to examine the effects of age, L2 proficiency, L1–L2 script and language distance, and measurement characteristics. The results showed that L2 grammar knowledge (r = .85), L2 vocabulary knowledge (r = .79), and L2 decoding (r = .56) were the three strongest correlates of L2 reading comprehension. The six low‐evidence correlates had moderate‐to‐strong mean correlations, with L2 listening comprehension being the strongest correlate (r = .77) and metacognition (r = .32) being the weakest correlate. Age, some measurement characteristics, and L1–L2 language distance were found to be significant moderators for some reading components.