A group of 25 honey bee venom allergic patients were treated with commercial honey bee venom at a monthly maintenance dose of 100 micrograms for approximately one year. At the end of one year 24 patients were intentionally challenged and one was accidentally challenged. Three patients experienced significant systemic reactions to challenge and three experienced minor reactions. Sera obtained before commencing therapy, at maintenance and before challenge were tested by radioallergosorbent test (RAST), double antibody technique and protein A RAST for IgE and IgG antibody levels to all five known honey bee venom allergens. All of the treatment failures experienced at least a two-fold rise in IgG antibody against phospholipase. The ratios of IgG to IgE antibodies in the pre-challenge specimens were analyzed by a graphical method. Four patients had inadequate responses to at least three of the five allergens and three of these patients were those who experienced severe reactions to challenge. Sixteen patients had adequate responses to all five allergens, four patients to four allergens and one patient to three allergens; three of these patients experienced minor or local reactions to challenge and the remainder no reaction. No single allergen identified only the three severe reactors but three allergens identified all three reactors. The diagnostic efficiency of the criteria used for assessing protection was 0.96. The only non-correlating case was classifying a single nonreactor as at risk. No patients were misclassified as protected.