We studied 55 subjects who had had anaphylactic reactions to bee stings within the previous 3 years. 38 out of 54 tested had IgE antibody to honey bee venom (HBV) as measured by radioallergosorbent test (RAST). On skin testing, 30 out of 34 had a positive test to HBV. Of these, 26 had a positive RAST. A positive skin test to HBV at high dilution or else a high anti-HBV RAST score appeared to identify those who, in a 6-month follow-up period, were at risk of developing further anaphylaxis following bee stings or immunotherapy. Of the two tests, RAST appeared to be the less sensitive. Measurements of IgG antibody to phospholipase A were seldom available for the period immediately preceding an anaphylactic episode and proved to be a poor means of predicting the liability to bee sting anaphylaxis in subsequent months.