The reproducibility of patch tests using allergens dispersed in petrolatum or water and placed into Finn chambers (Epitest Ltd, Oy, Hyryla, Finland) was studied in a double-blind fashion. A total of 128 patients were tested concomitantly to three different lots of an allergen to which they previously reacted positively and to one lot of the appropriate vehicle. The following allergens were evaluated with the total number of patients studied given in parentheses: benzocaine (10), mercaptobenzothiazole (9), cinnamic aldehyde (11), formaldehyde (14), nickel sulfate (15), quaternium-15 (15), neomycin (12), para-phenylenediamine (15), thiuram mix (14), and ethylenediamine dihydrochloride (13). Overall, the reproducibility of the triplicate responses among these patients was 88% (113/128). Reactions to benzocaine, mercaptobenzothiazole, thiuram mix, and nickel were reproducible in all patients. Reactions to cinnamic aldehyde (73%) and formaldehyde (79%) were the least reproducible. Eighteen percent (23/128) of patients failed to react to all of the allergen lots tested, suggesting “lost positive” reactions caused by either diminished sensitivity or a prior false-positive response. Among the allergies most likely to be lost were those for neomycin, formaldehyde, cinnamic aldehyde, thiuram mix, quaternium-15, and para-phenylenediamine. In contrast, reactions to benzocaine, nickel, mercaptobenzothiazole, and ethylenediamine were rarely lost. Failure to read the patch test sites later than 48 hours after application would result in a significant number of missed true-positive responses, especially for formaldehyde, quaternium-15, neomycin, and ethylenediamine dihydrochloride. These findings are discussed in light of prior published studies.