Two groups of birch pollen--allergic patients with seasonal rhinoconjunctivitis and asthma were followed during two consecutive birch-pollen seasons, one group, N = 10, during a season with high pollen load, and one group, N = 15, during a season of low pollen load. Half the patients were treated with immunotherapy (IT) for 3 and 4 years, respectively. The other half of the patients served as control group (non-IT). Bronchoalveolar lavage (BAL) was performed once before each season and once during the pollen season. Eosinophil (EOS) numbers in BAL were increased (p less than 0.01) during the season with high pollen load but not in the season with a low pollen load, and this increment was absent in the IT-treated group. Also, the EOS cationic protein levels were raised in the non-IT-treated group during the season with a high pollen load. The levels of EOS and neutrophil chemotactic activity were raised in BAL in both seasons in the non-IT-treated group compared with the IT-treated group (p less than 0.02, p less than 0.003, p less than 0.04, and p less than 0.005 in high- and low-load pollen season, respectively). Serum and BAL eosinophil chemotactic activity (ECA) were positively correlated (p less than 0.001). We conclude that there is an influx of active EOSs into the lung of pollen-allergic patients with asthma during a pollen season, which may be abrogated by IT. Furthermore, the generation of ECA appears to be an extremely sensitive marker of antigenic exposure, and the potent inhibition of the generation of ECA by IT may provide a clue as to the mechanism of this treatment.