Obsessive compulsive disorder (OCD) typically involves cycling between symptoms of intrusive aversive thoughts (obsessions) and repetitive rituals aimed at avoiding these aversive outcomes (compulsions), which interferes with patients' engagement with other important aspects of their lives. This cycling relationship between obsessions and compulsions highlights a potential role of impaired threat processing and avoidance behavior in OCD symptoms. The most effective behavioral therapy for OCD, exposure with response prevention (ERP), aims to break this cycle. However, it can be difficult for patients to access and engage with, suggesting a need for improved understanding of the neural mechanisms of threat processing and avoidance behavior in OCD to guide development of new and more effective treatments. Platform mediated avoidance (PMA) has proven to be a useful translational paradigm for use in rodents to examine avoidance neurobiology and models relevant to OCD, such as ERP and overtraining-induced persistent avoidance. However, to date this protocol has only been used in rats, and studies in transgenic mouse models relevant to OCD may shed further light on neural mechanisms relevant to disturbances in avoidance and threat processing in the disorder. To address this gap, we tested Sapap3 knockout (KO) mice, a leading preclinical model in OCD research, in the PMA task. Using this paradigm, we examined avoidance acquisition, expression, and extinction, as well as reward seeking under motivational conflict in two separate cohorts conditioned using higher (0.4 mA) or lower (0.24 mA) intensity shock. Surprisingly, the most striking difference observed in Sapap3-KOs vs control mice was heightened suppression of lever pressing for rewards during a tone signaling impending threat, suggesting a shift in action selection under motivational conflict (genotype effect avoidance conditioning: 0.24 mA cohort p=0.011, 0.4 mA cohort p=0.07; avoidance extinction: 0.24 mA p=0.057, 0.4 mA p=0.042). Avoidance responding was also acquired more slowly in Sapap3-KOs trained with a low intensity shock (time x genotype interaction p=0.025) and was extinguished more robustly following ERP (genotype effect p=0.043). In contrast, avoidance was similar between Sapap3 KOs and WT littermate controls trained using higher intensity shock (0.4 mA cohort time effect p<0.0001). Expression of the immediate early gene c-Fos associated with reinstatement of avoidance after ERP showed preliminary evidence for decreased activity of medial orbitofrontal cortex (mOFC) in KOs which may contribute to observed differences in PMA performance (KO vs WT mOFC c-Fos t-test p=0.0456). Together these findings suggest that mOFC dysfunction may contribute to increases in the influence of threats vs rewards over action selection in an animal model with relevance to OCD, and that the Sapap3-KO model presents valuable opportunities for deeper mechanistic investigation of avoidance and threat processing relevant to the human disorder.