Abstract Investigations into the experimental and observer variations of the Mantoux and the multiple puncture (Heaf) tests are described. They were carried out on patients in a mental hospital and on students in four high schools in Western Australia. Tests were performed by one operator and the reactions read by up to three observers on from one to five occasions. Old Tuberculin, human type PPD and Battey type PPD were used. When two identical Mantoux tests were read by one observer there was agreement between the recorded diameters of reactions within 3mm. in 85 % of cases. With two Heaf tests there was agreement in grading in 80 %,. With three observers reading each Mantoux test three times 80% of the readings, for each observed, were within 3mm. of each other and 90 % within 4mm. With the Heaf test read by three observers on five occasions there was complete agreement on grading in 65–75 %, when grades III and IV were combined. The variation between readers with each test was of a similar order to the variation within readers. The Heaf test appears to be only slightly less subject to observer variation than the Mantoux test. It seems reasonable in comparative Mantoux testing to accept a difference of 3 mm. or more as indicating differences in reaction to the sensitins used.