Patients with obsessive-compulsive disorder (OCD) display disrupted performance and abnormal lateral orbitofrontal cortex (LOFC) activity during reversal learning tasks, yet it is unknown whether compulsions and reversal learning deficits share a common neural substrate. To answer this question, we measured neural activity with in vivo calcium imaging in LOFC during compulsive grooming and reversal learning before and after fluoxetine treatment. Sapap3-knockout (KO) mice were used as a model for OCD-relevant behaviors. Sapap3-KOs and control littermates were injected with virus encoding GCaMP6f and implanted with gradient-index lenses to visualize LOFC activity using miniature microscopes. Grooming, reversal learning, and neural activity were measured pre- and post-fluoxetine treatment (18mg/kg, 4 weeks). Baseline compulsive grooming and reversal learning impairments in KOs improved after fluoxetine treatment. Additionally, KOs display distinct patterns of abnormal LOFC activity during grooming and reversal learning, both of which normalize after fluoxetine. Finally, reversal learning-associated neurons are distributed randomly amongst grooming-associated neurons (i.e. overlap is what would be expected by chance). In OCD, the LOFC is disrupted during both compulsive behaviors and reversal learning, yet whether these behaviors share common neural underpinnings is unknown. We find that the LOFC plays distinct roles in compulsive grooming and impaired reversal learning and their improvement with fluoxetine. These findings suggest that LOFC plays separate roles in pathophysiology and treatment of different perseverative behaviors in OCD.