This study examined (a) the effects of metacognition and proficiency on EFL reading performance and (b) the relation of metacognition and EFL reading performance. Data were collected by Metacognitive Awareness Inventory (MAI) and reading scores were examined. By variance analyses, we found that reading scores and metacognitive knowledge show variations across proficiency levels. There were no effects of time, timeXclass, and timeXproficiency on metacognition. When tests incorporated higher order thinking skills, participants' metacognitive knowledge or regulation correlated with reading scores, positively. Although trends that can explain differences in metacognition did not follow a pattern; it was observed that different proficiency groups benefited from training differently; for low- and mid-proficiency groups, a slight increase in metacognition regulation; and for high-proficiency group, a refinement in metacognitive knowledge was observed. We suggest instructional and assessment practices incorporate metacognition regarding learners' proficiency levels. Therefore, all students might see the relevance of metacognition and take responsibility for it.