Several aspects of a feature-matching theory formulated to account for the roles of stimulus novelty and significance in determining orienting responses (ORs) were examined. Stimulus significance and stimulus novelty were independently manipulated, and the electrodermal component of the OR was measured while sequences of compound pictorial or verbal stimuli were presented. Each sequence included a test stimulus, which was either significant or neutral, and was preceded by several standard control stimuli. Novelty was manipulated by either adding or deleting components from the standard stimuli, and significance was defined within the framework of the guilty knowledge paradigm. A change in the number of components of both verbal and pictorial test stimuli affected OR magnitude for significant but not for neutral test stimuli. Similar ORs were obtained regardless of the direction of stimulus change.