Prominent clinical models of chronic pain propose a fundamental role of classical conditioning in the development of pain-related disability. If classical conditioning is key to this process, then people with chronic pain may show a different response to pain-related conditioned stimuli than healthy control subjects. We set out to determine whether this is the case by undertaking a comprehensive and systematic review of the literature. To identify studies comparing classical conditioning between people with chronic pain and healthy control subjects, the databases MEDLINE, PsychINFO, PsychARTICLES, Scopus, and CINAHL were searched using key words and medical subject headings consistent with ‘classical conditioning’ and ‘pain.’ Articles were included when: 1) pain-free control and chronic pain groups were included, and 2) a differential classical conditioning design was used. The systematic search revealed 7 studies investigating differences in classical conditioning between people with chronic pain and healthy control participants. The included studies involved a total of 129 people with chronic pain (fibromyalgia syndrome, spinal pain, hand pain, irritable bowel syndrome), and 104 healthy control participants. Outcomes included indices of pain-related conditioning such as unconditioned stimulus (US) expectancy and contingency awareness, self-report and physiological measures of pain-related fear, evaluative judgements of conditioned stimulus pleasantness, and muscular and cortical responses. Because of variability in outcomes, meta-analyses included a maximum of 4 studies. People with chronic pain tended to show reduced differential learning and flatter generalization gradients with respect to US expectancy and fear-potentiated eyeblink startle responses. Some studies showed a propensity for greater muscular responses and perceptions of unpleasantness in response to pain-associated cues, relative to control cues.