Little is known from population-based studies in children about the diagnostic values of allergen-specific IgE antibodies (RAST) and skin prick test (SPT) with respect to hay fever. We aimed to determine and compare the diagnostic values of SPT and RAST to aeroallergens with respect to the incidence of hay fever cases in schoolchildren at different cut-off points. A prospective cohort study was performed on 1100 school children (5–7 and 8–10 years). Information on a doctor's diagnosis of hay fever was obtained by questionnaire and allergic sensitization to grass and birch pollen, cat, and Dermatophagoides pteronyssinus were measured using SPT and RAST between September 1992 and July 1993. Thirty-eight children give a history of hay fever (3.5%) in 1992/93 and additionally 37 cases occurred until 1996. Allergic sensitization was present in 17.9% (SPT), 30.2% (RAST) and more frequent in children with a history of hay fever (SPT: OR 11.7, 5.5–24.7; RAST: OR 10.6, 4.3–26.4). This difference was most pronounced for sensitization to pollen allergens. The sensitivity, specificity, positive and negative predictive values (PPV, NPV) for SPT and RAST were 65.6, 83.7, 11.9, 98.6 and 79.3, 71.6, 9.3, 99.0, respectively, with differences for specificity being significant (p < 0.001). Whereas NPV were equally high for SPT (99.2) and RAST (99.3), the incidence of hay fever cases were predicted rather poorly though somewhat better by SPT than by RAST (PPV 16.7 vs. 9.8; p < 0.001) initially. With increasing cut-off point for RAST reactivity, the PPV increased and reached 25.0 at 17.5 kU/l, whereas the NPV decreased to 97.9, which was lower than that of SPT reactivity (p < 0.01). At the cut-off point of 1.5 kU/l almost identical predictive values for SPT and RAST were obtained. SPT and RAST perform better in the negative than positive prediction of hay fever cases in epidemiological studies. Differences in the predictive capabilities depend on the chosen cut-off point for RAST reactivity.