"Kennel cough" in dogs in animal shelters is readily transmissible, reduces adoption rates, and commonly leads to the euthanasia of affected dogs. In cats, tracheobronchitis, conjunctivitis, and pneumonia have been associated with Bordetella bronchiseptica infection-but most cases of upper-respiratory infection (URI) probably are caused by herpesvirus and calicivirus, and many B. bronchiseptica culture-positive cats are clinically normal. Our prospective observational study was undertaken to document the contribution of B. bronchiseptica to disease in cats and dogs from two animal shelters undergoing outbreaks of canine kennel cough, to evaluate whether cross-species transmission might have occurred, and to determine if the presence of infected cats represented a risk to dogs. Clinically defined cases of kennel cough in dogs and URI in cats were investigated in two shelters by calculating clinical-disease incidence, alveolar-lavage cytological examination, bacterial and viral cultures, antibiotic-susceptibility testing, and molecular fingerprinting by pulsed-field gel electrophoresis. In a 40-cat and 40-dog "no-kill" shelter, the prevalences of culture positivity were 47% for B. bronchiseptica and 36% for calicivirus at the same time as two resident dogs demonstrated clinical cough. When no dogs had kennel cough 3 months later, 10% of cats were B. bronchiseptica-culture-positive and 63% calicivirus positive. In a large traditional shelter, the incidence of kennel cough in dogs increased over 12 weeks to a maximum of 19 cases/week/120 dogs, during which time the culture prevalence was 23% for B. bronchiseptica in dogs and 47% in cats. Three to 6 months before the kennel-cough epidemic, no dogs or cats were B. bronchiseptica positive. Very little genetic variability was detected in isolates from these shelters; all isolates except one corresponded to a single strain type which was identical to the pattern in a vaccine used in these shelters. Isolates from other cats, a horse, a llama, and a sea otter were genetically distinct from the shelter isolates. There was widespread resistance to cephalosporins and ampicillin, but low or no resistance to amoxicillin/clavulanate, trimethoprim-sulfamethoxazole, tetracycline, and enrofloxacin. Greater percent resistance was observed in the traditional shelter than in the no-kill shelter and feline isolates were more likely to be resistant than canine isolates.