To evaluate the activity of co-amoxiclav (amoxycillin/clavulanic acid) against Legionella pneumophila in vivo, a model of L. pneumophila pneumonia was developed in weanling rats rendered leukopenic by pre-administration of cyclophosphamide. Assessment of therapy was by lung bacterial counts and histological examination. Amoxycillin was ineffective in reducing bacterial counts in the lungs of infected rats, whereas erythromycin, the standard agent, was significantly more effective (P < 0.01). Co-amoxiclav and erythromycin, administered parenterally, produced significant bactericidal effects (P < 0.01), reducing the counts of L. pneumophila strain 1624 at 96 h to 1.2 log10 cfu/lungs compared with counts of 6 log10 cfu/lungs in the untreated animals. Clavulanic acid was also highly effective in preventing development of the infection, and was as efficacious as co-amoxiclav. Because of the significant reduction in bacterial numbers, a marked reduction in inflammation and consolidation of lung tissue was seen in rats treated with erythromycin, clavulanic acid or co-amoxiclav. The activity of co-amoxiclav was no greater than clavulanic acid alone, and no synergy was noted between the two components. When therapy was delayed until 48 h after infection, co-amoxiclav was as effective as erythromycin, with both treatments reducing bacterial numbers to 3.3 and 3.6 log10 cfu/lungs by 96 h, after only two days of therapy, in comparison with non-treated rats (5.6 log10 cfu/lungs). In a prolonged infection, produced by extending the period of leucopenia, co-amoxiclav and erythromycin were equally effective in preventing growth of the organism, with 1.5 and 1.6 log10 cfu/lungs, respectively, present at 96 h, in contrast to the non-treated rats with 5.7 log10 cfu/lungs (P < 0.01). After cessation of therapy, regrowth of L. pneumophila occurred in the erythromycin-treated group to such a degree that by 168 h, lung viable counts from these rats were significantly higher (4.8 log10 cfu/lungs) than in co-amoxiclav-treated rats (2.1 log10 cfu/lungs) (P < 0.05). Oral therapy of this infection with erythromycin or clavulanic acid, either alone or in combination with amoxycillin, resulted in counts of 3.3, 3.6 and 3.5 log10 cfu/lungs at 96 h, respectively. Although oral therapy was significantly less effective than parenteral therapy (P < 0.05), the bacterial counts in the treated groups were significantly lower than in the non-treated animals. The data show that co-amoxiclav displayed bactericidal activity consistently against intracellular L. pneumophila in vivo.