There is conflicting evidence regarding the reproducibility of patch testing. Discordant results have been reported in up to 44% of cases. The clinical relevance of these discordant patch tests has not been previously assessed. We studied 383 consecutive patients receiving simultaneous duplicate patch testing on opposite sides of the upper back with 10 allergens from the European standard series. Completely discordant patch tests-a negative test on one side with a positive test on the opposite side-were recorded in 30 (8%) patients. Two patients had discordant tests to two of the allergens; 28 had discordant reactions to one allergen. Completely discordant tests were recorded for nickel in 10 (3%) patients, balsam of Peru in two (0.5%), thiomersal in one (0.3%), cobalt in four (1%), paraphenylenediamine in three (0.8%), fragrance mix in two (0.5%), formaldehyde in four (1%), potassium dichromate in two (0.5%), lanolin in three (0.8%) and Kathon CG in one (0.3%). Of those patients with completely discordant patch tests, the allergen was deemed to be a true positive in 11 (3% of total) cases and of possible relevance in a further three. The allergen was felt to be relevant to the presenting complaint in seven (2% of total) patients.