Background Human models of migraine have been used for the past 30 years to test putative ‘trigger’ molecules and ascertain whether they induce migraine attacks in humans. However, nocebo effects using this model have never been systematically explored. Objective To assess the nocebo response rate in randomised clinical trials conducted at the Danish Headache Center, and in which human models of migraine were used. Methods In this systematic review and meta-analysis, we searched PubMed for studies of human models of migraine with a randomised, double-blind, placebo-controlled, two-way crossover design that included data on the incidence of migraine attacks or headache after infusion of placebo. A total of 943 articles were screened by title and abstract. Of these, 27 studies met the inclusion criteria (published between 1994 and 2020) and were included in the qualitative and quantitative analysis. We performed a random effects meta-analysis for the incidence of migraine attacks or delayed headache after placebo infusion. Results Twenty-seven studies were eligible for inclusion: 12 studies reported data for adults with migraine (n = 182), whereas 16 studies reported data on healthy volunteers (n = 210). For adults with migraine, the incidence of migraine attacks after placebo was 8.1% (95% CI = 2.5–15.5%, I2 = 50.8%). The incidence of delayed headache was 25.9% (95% CI = 18.5–34.1%, I2 = 18.9%). For healthy volunteers, the incidence of migraine attacks after placebo was 0.5% (95% CI = 0.0–3.6%, I2 = 0.0%) while the incidence of delayed headache was 10.5% (95% CI = 4.8–17.6%, I2 = 45.2%). Conclusion The nocebo response in randomised, placebo-controlled two-way crossover trials with intravenous infusions with intravenous infusion of placebo in migraine is negligible. Future studies using human models of migraine can be conducted by assuming a nocebo response rate of 15.5%.