New devices for puncture tests have been proposed recently, but their precision by comparison to the prick test method is poorly known. Seven puncture tests (Allerprick, Morrow Brown standardized needle, Phazer, Pricker, Stallerpointe, Stallerkit, and Wyeth bifucated needle) were compared with the modified prick test performed with hypodermic or intradermal needles in eight carefully selected normal volunteers. Skin tests with histamine hydrochloride (10 mg/ml) were only performed when there was no factor that might interfere with their interpretation. The site of skin tests on the forearm was demonstrated not to significantly influence the reaction size. The coefficient of variation of the tests ranged from 8.4% to 21.7%. Modified skin prick tests are satisfactory since they are highly reproducible (coefficient of variation: 13.4% and 16.5%) and there is no subject effect. Phazet was found to be more reproducible without subject effect. Pricker is satisfactory since it has no subject effect and a reproducibility similar to that of modified prick tests. Other tests are less reproducible (Stallerkit or Morrow Brown) or vary between subjects (Allerkit, Stallerkit, Stallerpointe, and Wyeth Needle).