This study investigates aspects of validity of an alternative measure of productive vocabulary. Lex30, developed by Meara and Fitzpatrick, is a word association task that claims to give an indication of productive vocabulary knowledge. Previous studies of Lex30 have assessed test-retest reliability, performance against native speaker norms, concurrent validity, reliability of parallel forms, and ability to reflect improvements in vocabulary development. In addition, the issue of construct validity has been explored. The study described here replicates some of these investigations with a different population and extends the investigation of construct validity. By comparing the performance of second language (L2) learners at different proficiency levels, the ability of the test to distinguish between levels of proficiency is explored. Concurrent validity is explored by comparing L2 learners' performance on Lex30 with that of two other productive vocabulary tests. Finally, one aspect of construct validity is explored by assessing whether Lex30 measures productive vocabulary use or simply recall. The findings indicate that Lex30 is a reliable and valid measure of productive vocabulary knowledge, but whether it measures only recall, or whether it measures actual ability to use vocabulary meaningfully and appropriately, appears to depend on the proficiency level of the test taker. © 2012 Taylor & Francis Group.