Alternative interpretations of startle probe modulation by a pictorial foreground were tested: Either reflex amplitude varies as a function of modality-determined attention allocation, or, regardless of probe modality, reflex amplitude varies with the emotional valence of the foreground content. Thirty-six subjects viewed a series of 54 slides, divided into two 27-slide blocks. Each block consisted of nine exemplars of three independently rated emotional content categories--pleasant, neutral, and unpleasant. Startle probes, half visual (flashgun) and half acoustic (white noise), were presented unpredictably during and between slide presentations. Eyeblink reflexes, corrugator and orbicularis oculi muscle tension, heart rate, and skin conductance were recorded during a 6-s slide interval. Subjects subsequently rated the slides for emotional valence and arousal, and interest value. Free-viewing times were also recorded. Analysis of reflex response and all ancillary measures supported the hypothesis that the primary determinant of startle modulation was the emotional valence of foreground content.