Dogs that participate in animal-assisted interventions (AAIs), often called ‘therapy dogs’, commonly interact with humans whose immune systems are not functioning optimally. The advisability of feeding raw meat (including poultry) to these animals remains a highly contentious issue, in spite of increasing evidence that raw meat is frequently contaminated with Salmonella. We set out to determine if consuming raw meat influences the risk of therapy dogs shedding Salmonella and other pathogens. Two hundred healthy therapy dogs from Ontario and Alberta were enrolled. Between May 2005 and November 2006, fecal specimens were collected from each dog every 2 months for 1 year, along with a log of places visited, antimicrobial use within the home, dog health status and diet. Specimens were cultured for Salmonella, methicillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant enterococci (VRE), extended-spectrum cephalosporinase (ESC) Escherichia coli and Clostridium difficile. Forty (20%) of the dogs were reported to have been fed raw meat at some point during the year. The incidence rate of Salmonella shedding in the raw meat-fed dogs was 0.61 cases/dog-year, compared with 0.08 cases/dog-year in dogs that were not fed raw meat (P < 0.001). Controlling for therapy dog group, the repeated measures, and pig ear consumption and diarrhoea in the 2 months prior to specimen submission, dogs that consumed raw meat were significantly more likely to test positive for Salmonellaat least once during the year than dogs that did not eat raw meat [odds ratio (OR) 22.7; 95% confidence interval (CI) 3.1–58.8; P < 0.001]. Specific Salmonella serovars were more common among dogs that consumed raw meat versus those that did not include S. Typhimurium, S. Heidelberg and S. Kentucky. Raw meat consumption was also significantly associated with shedding ESC E. coli (OR 17.2; 95% CI 9.4–32.3). No associations between C. difficile, MRSA or VRE and consumption of raw meat were detected. We recommend that dogs fed raw meat should be excluded from AAI programmes, particularly when the programmes involve interaction with humans at high risk of infection or adverse sequelae attributable to infection. Furthermore, although AAI dogs may not be representative of the general population of dogs, we also recommend that feeding of raw meat to dogs is to be avoided in homes where immunocompromised people live.