250 male students were assigned to 4 conditions of task predictability, 3 retention intervals, and 2 levels of training in a 4 X 3 X 2 design. Predictability was determined by irregular step-function tasks that differed in the proportions of systematically repeating (predictable) targets, ranging from fixed to random sequences. Integrated absolute error served as a performance criterion. In addition, 6 indexes of temporal-spatial patterning were obtained. Results showed greatest improvement and greatest absolute retention losses for the fixed task and a fixed-direction task added to the design. Intermediately predictable tasks did not differ in error from the random task; however, differences in response organization were found among all tasks. Results suggest the nature of changes in response organization, and indicate that timing may be most crucial for acquiring and maintaining skill. (24 ref.) (PsycINFO Database Record (c) 2006 APA, all rights reserved)