Recent evidence suggests that attentional capture is contingent on the attentional control setting induced by the task demands (C. L. Folk, R. Remington, & J. C. Johnston, 1992). Because the experiments on which these conclusions are based can be criticized for several reasons, the contingent capture hypothesis was tested using 2 visual search tasks in which subjects searched multielement displays in which a color singleton and onset singleton were simultaneously present. Both experiments show that the contingent capture hypothesis does not hold: Irrespective of attentional set, attention was captured by the most salient singleton. The findings suggest a stimulus-driven model of performance in which selection is basically determined by the properties of the featural singletons present in the visual field.