This paper reports an investigation into whether a test of productive vocabulary size using a category generation task can be useful and effective. A category generation task is a simple task where learners are asked to name as many as words as they can from a prescribed category such as "animals" or "body" parts. The virtue of this approach is that it potentially allows an estimate of productive vocabulary size, comparable to receptive size estimates, to be made. Four such tasks were trialled on 92 learners ranging from elementary to advanced level. Subjects also took Nation's Productive Vocabulary Levels Test (PVLT) (2001) and Meara & Milton's X-Lex (2003). The results suggest that category generation tasks can produce vocabulary size estimates and these are comparable in size with PVLT and about one third of the size of a receptive vocabulary size estimate (X-Lex). The tests appeared very reliable and can distinguish between learners of different levels of performance. There are still issues to be resolved concerning the tasks which can be used and the volumes of vocabulary they can potentially obtain. Factor analysis suggests the receptive and all the productive tasks test a single factor.