BACKGROUND Studies have shown that not all positive patch test reactions are reproducible upon retesting, that is, persistent. Non-persistent reactions might represent initial false-positive reactions, meaning that patients might unnecessarily avoid allergens. OBJECTIVE To investigate the occurrence of both persistent and non-persistent patch test reactions, to explore possible explanations, and to investigate whether allergen-specific differences exist. METHODS A retrospective analysis was performed on patients who were patch tested at least twice between 1 January 1995 and 31 October 2016, with at least one positive patch test reaction to an allergen that had been retested. Both univariable and multivariable analyses were performed to investigate the influence of several factors on persistence. RESULTS Of 274 retested positive reactions in 119 patients, 183 (66.8%) reactions remained positive. The strongest predictor for non-persistence in both univariable and multivariable analyses was strength of the first patch test, with weak positive reactions being significantly less persistent. Regarding allergen groups, metals and fragrances were less persistent than other allergens. CONCLUSIONS Weak positive reactions have a low persistence rate, and the dermatologist should be conservative in advising the patient on avoidance of these allergens, especially if clinical relevance is uncertain.