Several ways of structuring the early trials of a complex concept formation task were compared. Training trials were divided into two segments: (1) an asynchronous segment with one relevant and one irrelevant cue held constant while one relevant and one irrelevant cue varied and (2) a synchronous segment with all cues free to vary. The asynchronous segment was further divided into two parts in which (1) the more relevant cue was free to vary and (2) the less relevant cue was free to vary. The four possible orders of presenting the conditions were compared with each other and with a control group in which only synchronous segments were given. Eleven people in each of two experimental groups and one control group were given 160 presentations of the McHale-Stolurow (1962) task in five blocks of 32 trials. The first two blocks were training trials; the last three were the task trials. Performance on asynchronous trials was superior to performance on synchronous trials during training although no significant difference appeared on the task trials. There was no difference due to the ordering of asynchronous and synchronous blocks during training. During asynchronous trials, it proved better to present the more relevant cue varying first. (KB)