Twenty-six asthmatic children participated in a controlled trial of house dust mite avoidance. Group A (n = 14) had 12 weeks' avoidance, provided that they had not deteriorated at the end of the first 6 weeks. Group B (n = 12) had an initial 6 weeks' observation period followed by 6 weeks' avoidance. Mite numbers were extremely variable and often low. There was a significant fall in mite numbers in the first, but not the second, 6-week period in group A. In group B there was a significant fall in mite numbers in both the first (observation) and second (active avoidance) periods. Active avoidance produced highly significant falls in total serum IgE. There were no significant changes in IgE during the observation period. This impressive immunological effect was not associated with any changes in the radio-allergo-sorbent assay (RAST) to house dust mite, or symptom scores; peak expiratory flow rates or histamine induced bronchial reactivity. More rigorous avoidance procedures in more severely affected asthmatic patients warrants further investigation.