Background/aims: Sporadic reports on immediate and delayed cutaneous reactions to insulin detemir, a modern insulin analogue, have raised unsupported claims of allergy of type I, III and IV. The purpose of this experimental study using a provocative design was to elucidate the potential mechanisms behind such skin reactions. Material and methods: A total of 40 patients with type 1 diabetes or insulin-requiring type 2 diabetes, all naïve to insulin detemir, were injected on the thigh with 0.lmL of insulin detemir (Levemir ®) administered with an 8mm needle at three different depths, i.e. intradermal, subdermal and subcutaneously. Saline was injected as control. Any cutaneous reactions were assessed after 10 and 30min, after 24 and 48h and after 7 days. Histopathology of positive reactions on day 7 was obtained. The study was randomized, controlled, double-blinded, and conducted in accordance with ICH-GCP guidelines. Blood flow was recorded with the Periflux PF5010, and skin colour (a *) with the DSMII colorimeter. Results: Clinical reading, flowmetry and colorimetry consistently showed delayed reactions after intradermal insulin injection (35 of 40 patients reacted with mainly weak reactions, P<0.05), peaking after 48h, contrasting no special reaction immediately after injection, except for reactions attributed to needle trauma. A total of 22 patients reacted on subdermal injection and 21 on subcutaneous injection. Histopathology on day 7 from 22 reactions in 15 patients showed a consistent pattern of inflammation with eosinophilia as typically observed in adverse skin reactions to a variety of medicines. Reactions were interpreted as non-specific biologic responses to the insulin different from direct toxic actions and classical allergic reaction patterns. Only one person registered itch/discomfort. A prick test vs. histamine reference excluded insulin detemir to be a pharmacological histamine releaser. Thus, provocative testing with insulin detemir produced delayed skin reaction but no immediate reaction. Measurement of circulating insulin detemir-specific antibodies by RIA before and after 3 months showed no increase. Conclusion: Non-allergic delayed skin reactions from intradermal and, to a minor degree, subdermal and subcutaneous injections of insulin detemir were frequent in this experimental study and showed a consistent histology pattern of inflammation with eosinophilia. Immediate reactions were not produced. The reactions are unlikely to be specific for insulin detemir, and other insulins should be studied in a similar provocative design.