Learning EFL vocabulary in context is an essential process for the enrichment of the mental lexicon; however, past studies have shown that the level of success of contextualized learning is often different between groups with different levels of lexical proficiency. The present study explores the conditions under which a lower proficiency group can gain as much vocabulary knowledge as a higher proficiency group, focusing especially on the mental process of generating an image from reading contexts. The experiment was carefully designed to take into account (a) learners' lexical proficiency, (b) instruction on how to read the context, and (c) quality of provided context. A significant difference in scores on a context-cued word-recall task between the two proficiency groups was found only when the context was less imageable; when the groups were presented with more imageable contexts and instructed to try to imagine the situation described by the context (referred to as imagery instruction), the lower proficiency group's scores were equal to the higher proficiency group's. Based on the finding that imagery instruction did not reduce learning efficiency, no reason appears to exist to avoid vocabulary teaching that focuses on the meaning of a highly imageable context.