The "Mozart effect" refers to an increase in spatial reasoning performance following exposure to music composed by Mozart. Empirical tests of the effect have resulted in an inconsistent pattern of findings with some studies producing the effect and others failing to do so. The majority of the investigations have relied on paper-and-pencil tests. It is argued that in order for the effect to be of value in education, music needs to be shown to benefit performance in both laboratory and naturalistic settings. This study was a first attempt to examine performance employing a more naturalistic task. Participants repeatedly negotiated a series of computer-simulated rooms immediately after they listened to a sonata by Mozart or a piece by Philip Glass. Evidence of significant learning was observed in both conditions, but there was no difference between listening conditions. It is concluded that critical variables affecting the relationship between music and spatial reasoning ability are yet to be discovered. (Contains 2 tables.)