Background: The recent World Health Organization position paper on allergen immunotherapy states that local reactions to immunotherapy are not predictive of subsequent systemic reactions. Nevertheless, in clinical practice dose adjustment after local reactions continues to be recommended, presumably in an effort to prevent future systemic reactions. Objective: We sought to determine whether dose adjustment versus no adjustment for local reactions during allergen immunotherapy influences the occurrence of subsequent systemic reactions. Methods: In a single-site allergy clinic before October 1, 1997, local reactions after allergen vaccine injection resulted in adjustment of the subsequent dose. After October 1, 1997, no dose adjustments were made for immediate and late local reactions. For the same 9-month period before and after the change in local reaction dose-adjustment policy, systemic reaction rates were compared retrospectively. For individuals experiencing a systemic reaction, local reaction rates and local reactions immediately preceding a systemic reaction were also compared before and after the policy change. Results: Comparing the 9-month period (October 1996-June 1997) preceding the policy change and the 9 months (October 1997-June 1998) after the change in policy, the systemic reaction rates (0.80% and 1.01%, respectively) were not statistically different (P = .24). Among those experiencing a systemic reaction, the rate of local reactions was unchanged (7.3% and 4.7%, respectively; P = .07), and the rate of local reactions immediately preceding a systemic reaction did not increase (18.8% and 10.5%, respectively; P = .37). The sensitivity of a local reaction predicting a systemic reaction at the next immunotherapy dose was 15%. Conclusions: A local reaction is a very insensitive predictor for a subsequent systemic reaction at the next allergen vaccine dose. Dose adjustment for most local reactions is unnecessary and may delay therapy, increase costs, and put the patient at increased risk of dose administration errors. (J Allergy Clin Immunol 2000;106:840-3.)