This paper investigates the relationship between venom IgG levels and protection from stings. Venom-specific IgG antibody levels have been measured by radioimmunoassay in untreated wasp- (n= 38) and bee-allergic (n=16) patients presenting with systemic reactions to stings and in a sub-group of these (wasp= 15; bee = 9), before and after the initial course of venom immunotherapy (VIT). A history was taken of all reactions, the last systemic reaction being graded on a scale of 1–8 and of the number and timing of stings. In untreated patients venom IgG levels were much higher in bee-allergic patients (mean ± s.e. = 68.2 ± 7.1 % positive pool) than in the wasp group (27.1 ± 4.2%) (P< 0.05 Mann-Whitney U-test). There was a marked rise in venom IgG after the initial course of VIT in the wasp group (geometric mean and 95% confidence intervals = 40.5%, 28.8−54.3) but a much smaller rise in the bee group (15.3%, 6.6–24.1), with no overlap in the 95% confidence intervals. Bee patients, who were mainly beekeepers or their relatives, had been more heavily immunized with venom than wasp patients. They had received: (i) more stings (mean number of stings: bee, 26; wasp, 4; P< 0.001) and (ii) more stings per year. Wasp patients received their smaller number of stings over a much longer period, up to 40 yr. There was no correlation between the severity of the last systemic reaction and the venom IgG levels alone or venom IgG and IgE levels in combined analysis in either bee or wasp patients. This study shows that the pattern of IgG response differs in bee and wasp-allergic subjects, and that most bee-allergic subjects with systemic reactions have high levels of venom IgG. The degree of immunization with venom seems to be an important determinant of the venom IgG level. Our findings suggest thai venom-specific IgG levels do not predict systemic reactions to stings and are not useful for monitoring VIT. If protection from stings is IgG-mediated, our observations suggest that the relevant immune response is more complex, possibly involving IgG sub-classes, IgG antibodies to individual venom antigens or antibody affinity, and not adequately reflected by measurement of the concentration of venom-specific IgG.