This study examined how well second language (L2) recall and recognition vocabulary tests correlated with a reading test, how well each vocabulary test discriminated between reading proficiency levels, and how accurate each test was in predicting reading proficiency when compared with corpus studies. A total of 116 college-level learners of English as a foreign language took a reading test and 2 vocabulary size tests: meaning recall and meaning recognition. Participants were divided into 4 reading proficiency levels based on the reading scores. We correlated the reading scores with the 2 vocabulary scores, compared the 4 reading groups on each vocabulary test, and compared the vocabulary size of each of the reading proficiency groups with corpus studies. Both vocabulary tests were good predictors of reading, but the recognition test fared slightly better. We introduce the notion of 'comprehension vocabulary' and suggest that a recall test is more appropriate for measuring sight vocabulary while a recognition test is more appropriate for measuring comprehension vocabulary.