Three experiments were undertaken to examine the processing limitations that underlie multiple task performance and, ultimately, to test the theory that there exists a time-sharing ability which is general and can be assessed through the simultaneous presentation of any two or more sufficiently demanding tasks. Subjects were males and/or females chosen from the University of Oregon paid subject pool. Analysis of variance and Fishers least significant difference test were used to assess results. It was concluded that time-sharing is not a single general ability, but rather is dependent upon several more specific, and perhaps independent, processing limitations. These include: (1) an inability early in practice to simultaneously select, or retrieve, multiple responses from memory; (2) a persisting inability to initiate multiple independent responses simultaneously; (3) an inability to efficiently process contiguous inputs from separate modalities owing to the need for a modality-specific attentional focus; and (4) an inability to efficiently process multiple inputs from within the same modality owing to the existence of structural interference. It is suggested that the prediction of performance on complex criterion task combinations requires specification of which of these component abilities is required by the criterion situations, and the tailoring or predictor tasks based on this specification. (Author/CWM)