Concern has been expressed that clavulanate can antagonize ticarcillin against enterobacteria and pseudomonads that have inducible expression of chromosomal 'Class I' beta-lactamases. It is suggested that clavulanate-induced enzyme inactivates ticarcillin, which itself is a feeble inducer. We confirmed that this mechanism applied, showing that antagonism was abolished in beta-lactamase-basal mutants of inducible strains. Antagonism has been reported in double disc tests with strains of Pseudomonas aeruginosa, Enterobacter cloacae, Citrobacter freundii, Serratia spp. and indole-positive Proteeae. Only with some strains of Ent. cloacae and Morganella morganii, however, did the presence of 1-32 mg/l clavulanate elevate the MIC of ticarcillin by more than one or two dilutions in chequerboard studies. Clavulanate was synergistic with ticarcillin against Proteus vulgaris strains, being a potent inhibitor of the unusual Class I enzyme of this species. Induction-determined antagonism was not reduced in Ent. cloacae transconjugants that produced the plasmid-mediated TEM-1 beta-lactamase, despite the ability of this enzyme to bind clavulanate. Our results suggest that Ent. cloacae and M. morganii strains should be confirmed not to be more sensitive to ticarcillin alone than to ticarcillin/clavulanate, before the latter combination is used clinically. Otherwise, it appears that beta-lactamase induction is unlikely to cause significant antagonism. It is emphasized that induction is reversible, causing, at worst, a transient resistance. It should not be confused with the selection of stably-derepressed mutants that can occur, for example, in the clinical use of newer cephalosporins.