Background: Although non-steroidal anti-inflammatory drug hypersensitivity (NSAID-H) has been widely studied in adults, there is still a lack of data regarding the features and phenotypes of NSAID-H in children. Our aim was to define risk factors and different phenotypes according to clinical patterns. Methods: Patients with a history of reaction to any NSAIDs referred between January 2012 and October 2014 were included. After completing a European Network for Drug Allergy (ENDA) questionnaire, initial skin and/or oral provocation tests (OPTs) were performed for the offending drug. Additional OPTs were done with aspirin in case of NSAID-H to determine cross-reactivity. NSAID-hypersensitive patients were defined as being either a selective responder (SR) or cross-intolerant (CI) and further categorized according to either the ENDA/GA2LEN classification or an alternative scheme by Caimmi et al. [Int Arch Allergy Immunol 2012;159:306-312]. Results: Among 121 patients [58.7% male, average age 7.8 years (4.7-10.8)] with 161 NSAID-related reactions, 110 patients with 148 reactions were assessed. NSAID-H was diagnosed in 30 (27%) patients with 37 (25%) reactions. Multivariate regression analysis revealed that an immediate-type reaction and respiratory symptoms during the reaction increased the risk of a reproducible NSAID-related reaction (OR 3.508, 95% CI 1.42-8.7, p = 0.007; OR 3.951, 95% CI 1.33-11.77, p = 0.014, respectively). Additional OPTs revealed 13 SRs and 14 CIs. A family history of allergic disease was more frequent in CIs compared to SRs (57.1 vs. 15.4%, p = 0.031). Reactions belonging to CIs were more frequently characterized by angioedema compared to those of SRs (81.3 vs. 46.2%, p = 0.019). SRs and CIs were further classified as single NSAID-induced urticaria/angioedema and/or anaphylaxis (n = 13), NSAID-induced urticaria/angioedema (n = 7), NSAID-exacerbated cutaneous disease (n = 2) and NSAID-exacerbated respiratory disease (n = 1). Four CIs could not be categorized according to either classification system. One SR could not be categorized according to ENDA/GA2LEN. Conclusion: During childhood, NSAID-H exhibits different phenotypes and the majority of them can be categorized with current classification systems; however, classifications based on adult data may not exactly fit NSAID-H in paediatric patients.