Beyond specific conflicts between tasks that are obviously similar (e.g. two verbal tasks) and limits specific to speeded responses, is there a general limitation on what tasks can be done simultaneously? In two experiments, we examined dual-task combinations designed to avoid known sources of specific interference. Under these circumstances, a general factor model predicts consistency in the pattern of results. Tasks should be ordered in demands on the general factor as measured by interference with concurrent tasks; this order should be the same for any concurrent task used to measure it. This prediction was confirmed in both experiments, each involving 12 dual task combinations of four tasks. In Experiment 1, the tasks were tone discrimination, random letter generation, a manual- tactile manipulation task,and recognition memory for photographs. In Experiment 2, the ® rst of these was replaced by an easier tone-monitoring task, and the last by a visual prototype learning task