A simulation game was designed to examine the impact of a guaranteed annual income (GAI). The sample of 158 player-objects included factory and clerical workers, high school dropouts, single mothers, and CEGEP and university social work students. To establish the validity of the game, the working and spending behaviour of players was compared with results reported for the New Jersey negative income tax experiment, and found to be similar in a number of respects. The game also simulated two features not present in the New Jersey experiment: (a) variable labour-market conditions, and (b) comparison of a partial, time-limited GAI with a permanent, universal plan. For players on a GAI, working hours were significantly lower when fellow-players were not on a GAI than when they were. Results suggest that work effort may be related to comparisions with a reference group on visible consumer goods.