Background. Clinical scenarios of repeated pain usually involve both nociceptive and non-nociceptive input. It is likely that associations between these stimuli are learned over time. Such learning may underlie subsequent amplification of pain, or evocation of pain in the absence of nociception. Methods. We undertook a systematic review and meta-analysis to evaluate the evidence that allodynia or hyperalgesia can be a classically conditioned response. A sensitive search of the literature covered Medline, Embase, CINAHL, AMED, PubMed, Scopus, PsycArticles, PsycINFO, Cochrane Library, and Web of Science. Additional studies were identified by contacting experts and searching published reviews. Two reviewers independently assessed studies for inclusion, evaluated risk of bias, and extracted data. Studies were included if they aimed to elicit or amplify pain using a classical conditioning procedure in healthy, adult humans. Studies were excluded if they did not distinguish between classical conditioning and explicit verbal suggestion as learning sources, or did not use experiential learning. Results. Thirteen studies, with varying risk of bias, were included. Ten studies evaluated classically conditioned hyperalgesia: nine found hyperalgesia; one did not. Pooled effects (n58 with full data) showed a significant pain increase after conditioning (mean difference of 7.40 [95%CI: 4.00–10.80] on a 0–100 pain scale). Three studies evaluated conditioned allodynia and found conflicting results. Conclusion. The existing literature suggests that classical conditioning can amplify pain. No conclusions can be drawn about whether or not classical conditioning can elicit pain. Rigorous experimental conditioning studies with nociceptive unconditioned stimuli are needed to fill this gap in knowledge.