Opioid use disorder (OUD) has been linked to exaggerated attentional, affective, and arousal responses to opioid-related stimuli, as well as altered responses to other affective (eg, naturally rewarding or aversive) stimuli, particularly blunted responses to pleasant/rewarding stimuli. Both exaggerated responses to drug-related stimuli and reduced response to pleasant stimuli may influence the course of OUD and its treatment, however interpretation of studies thus far is limited by methodological issues. In the present study, we examined subjective ratings, and attenuation of the P3 component of the acoustic startle-evoked event-related potential (as a measure of attention), while viewing neutral, pleasant, unpleasant, and drug-related images. Participants prescribed opioid agonist treatment (OAT) for OUD (n = 82) were compared to a carefully-matched control group (n = 33) and to recently-abstinent participants with OUD (n = 22). Relative to controls, participants prescribed OAT gave higher positive valence ratings of drug images, and blunted valence responses to other affective images, but groups did not differ in terms of arousal ratings or P3 amplitude. Within the OAT group, linear modeling of associations between frequency of recent illicit opioid use and startle P3 amplitude found an association between increased recent illicit opioid use and reduced attention to pleasant, relative to unpleasant, images. The latter finding may have implications for interventions targeting cognitive biases in people with substance use disorder. In particular, they suggest that enhancing attention to pleasant stimuli may be as, if not more important, than the typical approach of trying to reduce attentional bias to drug-related stimuli.