As part of a larger epidemiological study, 114 children with respiratory symptoms, born between 1978 and 1980, were skin-prick tested to Dermatophagoides pteronyssinus (DP), mixed grass pollens (G) and cat dander (C), and to histamine and saline controls (Bencard, U.K.) using 1 mm prick-lancets (Dome/Hollister-Stier), between July and September 1987 and again in October 1989. A weal > or = 2 mm to one or more allergens was regarded as a positive result. Each child was tested by the same investigator on each occasion, using similar techniques. Three children were excluded from analysis as they had failed to respond to histamine testing on one of the two occasions. In 1987, of the 111 children analysed, 58 (52%) children were skin-test positive, and 53 (48%) skin-test negative, while in 1989 62 (56%) were positive and 49 (44%) negative. Twelve children (11%) changed status from negative to positive, while eight (7%) changed from positive to negative. For the group as a whole the percentage agreement between the results obtained 2 years apart was 82%. In comparison to previous studies a greater number of subjects in this population than expected changed atopic status. We therefore further examined the data from those who had changed status and classified as borderline those subjects with no difference in weal size of greater than 2 mm for any allergen between 1987 and 1989. Only five children then changed status from negative to positive, none from positive to negative and 15 demonstrated only borderline changes. The coefficients of repeatability for the 106 children who did not change status were 3.37 mm, 2.80 mm and 2.33 mm for D. pteronyssinus, mixed grass pollens and cat dander respectively. The good short-term repeatability of the testing method was demonstrated in a group of 29 similar children; the coefficients of repeatability were 0.38 mm for DP and G, and 0.72 mm for C. These data demonstrate that, in a population of children with respiratory symptoms, skin-prick testing within individuals is highly repeatable over the short term, but poorly repeatable over a 2 year period. However, the percentage agreement in skin-prick test status for the group as a whole was high (82%). While no child became unequivocally skin-test negative having been previously positive, a small number of children changed status from negative to unequivocally positive, suggesting a genuine but small (4%) increase in the prevalence of skin-test positivity in this population.