BACKGROUND: This systematic review and meta-analysis evaluated whether people with obsessive-compulsive disorder (OCD) exhibit differences in associative learning. METHODS: CINAHL, Cochrane Library, EMBASE, PsycINFO, PubMed, Scopus and Web of Science were searched for published peer-reviewed studies in English quantitatively examining the relationship between OCD and associative phenomena in humans. 5508 titles, 124 abstracts and 55 full texts were reviewed; citation searching identified 15 records. Sixty-six studies were included. Risk of bias was assessed and random-effects meta-analysis synthesised results. RESULTS: Obsessive-compulsive disorder was associated with differences in extinction (g = 0.37, p < .001, k = 15) and mediated associative learning (k = 1), but not associative learning (g = -0.12, p = .26, k = 49), avoidance learning (g = 1.5, p = .13, k = 4), blocking (k = 1), generalisation (g = -0.2, p = .16, k = 3), latent inhibition (g = 0.45, p = .39, k = 3), outcome devaluation (g = 0.26, p = .33, k = 7), Pavlovian-to-instrumental transfer (g = -0.24, p = .22, k = 2) or reversal learning (g = 0.16, p = .36, k = 10). The quality of evidence was moderate for reversal learning, low for associative learning, extinction, generalisation, latent inhibition and Pavlovian-to-instrumental transfer, and very low for avoidance learning and outcome devaluation. CONCLUSION: Low quality evidence suggests people with OCD learn that stimuli no longer predict negatively valanced outcomes slower than healthy controls. Future research must elucidate the cause of attenuated extinction and its specificity to OCD.