Frogs can be useful models for studying the mechanisms that may regulate their natural microbial flora. Their skin glands produce a secretion containing 20-30 different peptides, some antimicrobial some neurotrophic. As they often live in soil or silt that is rich in microbes, they can be expected to be able to prevent or eliminate infections in very short peroids of time. The bacterium Aeromonas hydrophila is widely distributed in nature and is considered as part of the natural flora of frogs and many animals, including humans. From an alternative frog strain of A. hydrophila, Bo-3, we isolated a spontaneous and stable mutant (Bo-3N), resistant to nalidixic acid, here used to follow the host-microbe interactions in experimental infection of mouth and skin of Rana esculenta. The skin peptides had been previously isolated, sequenced and cloned. We showed that skin treatment with a glucocorticoid (GC) cream blocked de novo synthesis of these peptides and, simultaneously, prepropeptide mRNAs disappeared while IκBα was up-regulated. Experimental mouth infections with 20 million cells of A. hydrophila Bo-3N showed that a normal wild frog can eliminate the bacteria from the mouth within 15 min, while a frog pretreated with GC cream for 1 h could not reduce Bo-3N below 3500 colony-forming units (CFU)/5 µl 'saliva'. An in vitro comparison showed that frog blood or serum allowed bacteria to grow, while the skin secretion killed the bacteria within 10 min. Using different enzyme-linked immunosorbent assays (ELISAs) with rabbit anti-Bo-3 serum as a positive control, we were able to rule out immunoglobulin G (IgG) binding to A. hydrophila. An assay for immunoglobulin M (IgM) (or some other serum component) in frog serum showed binding to A. hydrophila only corresponding to a few per cent of the positive control. For skin infections we bathed the frogs for 10 min in an overnight culture of Bo-3N diluted to about 107 CFU/ml. Electrical stimulation after the bath showed, for the total secretion, a two to fourfold increase in the antibacterial activity, while a pretreatment with GC cream reduced the activity to about one-third of that of the non-bathed control frog. HPLC analysis of the peptide pattern confirmed these findings. The survival value of antimicrobial peptides have earlier been demonstrated in vivo and in vitro only in Drosophila. The present experiments are the first combined in vivo and in vitro demonstrations of the function of peptide antibiotics in a vertebrate. One such function is involved in the control of the natural flora.