-The present study was undertaken to investigate to what extent pregnancy diagnoses made by transrectal ultrasonography could be confirmed by measurements of plasma concentration of ovine pregnancy-associated glycoproteins (ovPAG). A total of 424 Awassi × Merino ewes were synchronized for estrus and examined by transrectal ultrasonography. In Experiment 1, the ewes (n = 156) were repeatedly scanned in a standing position on d 29, 36 and 50 of gestation. Similarly, the ewes (n = 268) in Experiment 2 were scanned on d 24, 29 and 34 of gestation, but these ewes were fasted for 12 h prior to the examination and the abdominal wall of each animal was lifted up by the hands of the assistant during the scanning. Blood samples were withdrawn after each transrectal ultrasonographic examination in both experiments. Ovine PAG concentrations were measured in plasma by a heterologous radioimmunoassay and the cut-off value for pregnancy was ≥ 1 ng·mL -1 . Based on the lambing performance, in Experiment 1, altogether 47 false negative and 38 false positive diagnoses were made by transrectal ultrasonography in 24 and 33 ewes, respectively between d 29 and 50 of gestation. In Experiment 2, altogether 8 false negative and 13 false positive diagnoses both were made in 7 ewes between d 24 and 34 of gestation. In both experiments, all ewes with false negative diagnoses had ovPAG concentrations higher than the threshold level for pregnancy diagnosis and all ewes with false positive diagnoses had ovPAG concentrations lower than the threshold of pregnancy. Furthermore, by the PAG-RIA test all lambed or aborted ewes (n = 63) were correctly diagnosed as pregnant and with three exceptions, all non-lambed ewes (n = 361) were correctly diagnosed as non-pregnant during the examined periods of both experiments.transrectal ultrasonography / ovPAG / pregnancy diagnosis / sheep Reprod. Nutr. Dev. 43 (2003) 577-586 577