A kinetic killing-curve method, designed to mimic several aspects of clinical therapy in endocarditis, was used to test 10 strains of Haemophilus parainfluenzae against 28 antibiotic regimens. In an effort to simulate changing in vivo levels of antibiotic in serum, concentrations of three penicillins, three cephalosporins, gentamicin, and chloramphenicol were sequentially adjusted over a 12-hr period. Against six beta-lactamase-negative strains, gentamicin in combination with penicillin or cephalosporin invariably resulted in an additive or synergistic effect. Chloramphenicol and a penicillin or cephalosporin usually displayed an indifferent effect, but chloramphenicol was often antagonistic when combined with gentamicin. With four beta-lactamase-positive strains, variable responses were noted to penicillin-aminoglycoside combinations; cephalosporin-aminoglycoside combinations were usually synergistic. This dynamic approach to killing-curve studies may be more appropriate than a static system for in vitro examination of the effect of antimicrobial combinations against selected organisms.