A study investigated the instructional usefulness of HANGMAN, a computerized vocabulary development game, for university students of English as a Second Language. In the ten microcomputers available to students, the regular version of the game was replaced with a modified version in which all student key-strokes were recorded automatically. Subjects were anonymous, most likely first-year university students. Data from 790 problems in 100 student sessions were analyzed for use of linguistic-competency-based (CB) and non-competency-based (NCB) strategies for solving the vocabulary puzzles presented. CB strategies included solving a puzzle with no hints; using hints with the result of a correct answer; and in some cases, use of a hint or "See Solution" to avoid hanging. NCB strategies included use of "See Solution" rather than persevering; abuse of hints; and patterned key-presses. Analysis indicates students used CB strategies to solve about half the words presented to them, while ESL instructors, viewed as "ideal" learners, used CB strategies to solve 92% of the problems presented to them. The students' manner of doing the problems was distinctive, with NCB strategies used at a much higher rate among this group. Recommendations are made for improving the program's instructional effectiveness. (MSE)