An epidemiological study was conducted to assess the rates of condemnation and the diagnoses made on condemned broiler chicken carcasses in a federally-inspected abattoir. The first objective was to determine the predictive value of a positive test: the proportion of birds that were truly unfit for human consumption among condemned carcasses. The second objective was to assess the degree of agreement between diagnoses made in the abattoir and diagnoses determined by a detailed gross postmortem examination on the same condemned carcasses. A two-stage convenience sampling technique was used to obtain the birds needed. Fifteen lots of birds were selected and within each of these, approximately 45 condemned carcasses were selected for a total of 680. All the diagnoses made on these birds at the abattoir were recorded and the carcasses individually identified. The sampled carcasses were transported to the Ontario Veterinary College where a panel of three federal veterinarians independently judged whether the birds were fit for human consumption. In addition, a detailed gross postmortem examination was conducted on each carcass. An overall predictive value of a positive test of approximately 70% was found. The degree of agreement among members of the panel of examiners was also tested. The overall agreement was good (Kappa = 0.62 between examiners 1 and 2 and 0.51 between examiners 1 and 3). However, carcasses condemned for specific conditions at the abattoir, namely valgus varus deformity, mutilation, cellulitis and hepatitis showed no agreement beyond chance, while other carcasses condemned for conditions such as cyanosis and bruising showed poor agreement concerning their disposition for at least one pair of examiners.(ABSTRACT TRUNCATED AT 250 WORDS)