Skin testing with the suspected compound has been reported to be helpful in determining the cause of cutaneous adverse drug reactions (ADRs), but the value and specificity of these tests need to be determined. In this study, 72 patients with presumed drug eruptions (27 maculopapular, 18 urticarial, seven erythrodermic, nine eczematous, four photosensitivity, three fixed drug eruptions, three with pruritus and one with acute generalized exanthematous pustulosis) were assessed. All had drug patch tests; 46 also had prick tests and 30 had intradermal tests (performed on hospitalized patients using a sterile solution of the suspected drug, diluted sequentially) with immediate and delayed readings. Among these patients, 52 (72%) had a positive skin test reaction, 43%, 24% and 67% in patch, prick and intradermal skin tests, respectively. The results of skin tests varied with the drug tested and with the clinical type of cutaneous ADR, as a significantly higher number of positive patch tests was observed in maculopapular rashes than in urticarial reactions (P = 0.001). This study supports the value of careful sequential drug skin testing in establishing the cause of cutaneous ADR. Guidelines are proposed for performing these tests, and these include the use of appropriate negative control patients to avoid false-positive results.