Endogenous Pain Modulation (EPM) impairment is a significant contributor to chronic pain. Conditioned pain modulation (CPM) testing assesses EPM function. Osteoarthritic (OA) dogs are good translational models, but CPM has not been explored. Our aim was to assess EPM impairment in OA dogs compared to controls using CPM. We hypothesized that CPM testing would demonstrate EPM impairment in OA dogs compared to controls. Dogs with stifle/hip OA and demographically-matched controls were recruited. The pre-conditioning test stimulus, using mechanical/thermal quantitative sensory testing (MQST or TQST), were performed at the metatarsus. A 22N blunt probe (conditioning stimulus) was applied to the contralateral antebrachium for 2 minutes, followed by MQST or TQST (post-conditioning test stimulus). The threshold changes from pre to post-conditioning (∆MQST and ∆TQST) were compared between OA and control dogs. Twenty-four client-owned dogs (OA, n = 11; controls, n = 13) were recruited. The ∆MQST(p < 0.001) and ∆TQST(p < 0.001) increased in control dogs but not OA dogs (∆MQST p = 0.65; ∆TQST p = 0.76). Both ∆MQST(p < 0.001) and ∆TQST(p < 0.001) were different between the OA and control groups. These are the first data showing that EPM impairment is associated with canine OA pain. The spontaneous OA dog model may be used to test drugs that normalize EPM function.