All four subclasses of IgG antibodies to timothy grass pollen extract were measured by a three-layer immunoradiometric assay in sera from 20 grass pollen-allergic patients who underwent specific immunotherapy in a 3-year prospective study. Both IgG1 and antibody level passed its peak (median 5,4 U/ml) after 12 weeks. At this time, the ratio between the medians of IgG1 and IgG4 antibodies was 2.25 IgG4 antibody level reached its peak (median 11.6 U/ml) just before termination of immunotherapy. At this time IgG1/IgG4 ratio was 0.43. Two years after the end of immunotherapy, IgG1 and IgG4 antibody levels were 0.0 and 1.8 U/ml in median, respectively. The amounts of IgG2 and IgG3 antibodies detected in the sera were less than 1.6 U/ml and were considerd insignificant. Preseasonal serum IgG1 and IgG4 antibody levels did not correlate significantly with symptom scores in the subsequent season. Serum IgG4 level obtained after 12 weeks of immunotherapy was significantly correlated to symptom score in the third season i.e. the season just after termination of therapy (rs =0.529. t= 2.567. P= 0.02). In this work. A scrum IgG4 antibody level higher than 8.0 U/ml after 12 weeks of therapy predicted poor clinical result at the end of immunotherapy with 100% sensitivity and 87% specificity. An IgG4/IgG1 ratio greater than 1.0 after 12 weeks' therapy had the same predictive value.