Abstract. Monitoring of Toxoplasma infection in animals destined for human consumption is a great challenge for human toxoplasmosis prevention. This study aimed to compare results obtained from a naturally infected population of sheep using different tests and targeting an original matrix: meat samples and muscle fluids collected at the slaughterhouse. A commercial ELISA test was performed on diaphragm fluids from 419 ovine carcasses collected at the slaughterhouse. A MAT (modified agglutination test) was performed on heart fluids obtained from the same animals. In addition, all hearts were bioassayed in mice. Serological test agreement, the relative sensitivity of ELISA MAT and mouse bioassay as well as a correlation between titres and parasite isolation probability were statistically evaluated. The overall agreement (kappa coefficient = 0.64) of ELISA on diaphragm fluids and MAT on heart fluids is substantial and subsequently both tests can be used for epidemiological studies. Relative sensitivity was higher for MAT performed on cardiac fluids (90%) than ELISA on diaphragm fluid (61%). For both serological tests, relative sensitivity is lower in lambs younger than 12 months. Relative sensitivity of mouse inoculation was 42%. A significant correlation was obtained between increasing MAT titres and probability to isolate live parasite from the heart. When the fluid titre was higher than 1:16, parasites were isolated in 65% of cases. When it was lower, isolation failed in 95% of the cases. According to our results, cardiac fluids appear to be a relevant matrix for toxoplasmosis survey in meat.