The hypothesis was advanced that the aspect of learning which makes performance of a task more compatible with other concurrent activities is not adequately reflected by any of the conventional measures of learning. This aspect of learning, it is proposed, is made possible in part by the degree of repetitiveness of the task, and can be measured by requiring simultaneous performance of a second task at various levels of practice on the primary task. The Ss were trained in either a random or a repetitive version of a visual-motor task which required anticipation and precise timing. A mental arithmetic task was added either early or late in practice on the visual-motor task. Arithmetic performance was comparable for the random and repetitive groups if the secondary task occurred early in practice. However, arithmetic scores were significantly superior for the repetitive group when arithmetic was added late in practice. Scores on the motor task remained comparable whether the secondary task was added early or late. These results indicate that extra-task performance on a mental arithmetic task can be used to measure an aspect of learning of a visual-motor task which is not reflected by the scores on the motor task itself, and that this aspect of learning is dependent upon the repetitiveness of the task.