Two studies were conducted to examine the time course of attentional and emotional processing using the startle eyeblink reflex. Forty-eight participants listened to a series of 1500-ms tones that occurred every 6 s. The tones signaled participants to generate emotional images that were positive or negative in valence and high or low in arousal. Auditory startle probes occurred 120 ms, 1400 ms, or 4000 ms after tone onset. Startle inhibition was seen 120 ms after tone onset and startle facilitation was found at 1400 ms, compared to startles elicited 4000 ms after tone onset. Startle inhibition was greater at 120 ms when the tone signaled imagery, indicating an attentional effect. A second experiment found that this effect was not caused by the comparative rarity of the tones signaling imagery. Startle magnitude was also smaller at 1400 ms when the tone signaled imagery compared to no-imagery tones. The type of imagery did not modulate the startle response 120 ms after tone onset, but negative valence imagery enhanced startle magnitude at 1400 ms and 4000 ms after tone onset, and high arousal also enhanced startle magnitude at 4000 ms. Thus, attention and emotion followed different time courses in affecting the startle reflex response during imagery.