Bovine cysticercosis, an infection in cattle caused by the zoonotic tapeworm Taenia saginata, leads to considerable global economic losses. The reported prevalence of bovine cysticercosis in Ethiopia has, until now, been underestimated due to the low sensitivity of meat inspection. This study aimed to estimate the true prevalence of bovine cysticercosis in two Ethiopian abattoirs, and evaluate the diagnostic performance of routine meat inspection, predilection-site dissection and Antigen-ELISA. A cross-sectional study was conducted and a Bayesian Latent Class Model was used to estimate the true prevalence and the diagnostic performance of the tests. The apparent prevalence of bovine cysticercosis was 2 %, 15 % and 14 % using meat inspection, predilection site dissection and Ag-ELISA, respectively. The estimated true prevalence was 39.0 % (95 %CI: 16.3, 93.7 %) in Jimma, and 31.7 % (95 %CI, 12.1, 77.9 %) in Ambo abattoirs. Predilection site dissection had moderate sensitivity (50.4 %, CI, 19.7, 96.4 %) and while Ag-ELISA had a lower sensitivity (17.3 %, CI, 5.0, 37.3 %). The sensitivity of Ag-ELISA for the detection of viable cysticerci only was 46.7 % (CI, 4.1, 91.1 %). Meat inspection had a much lower sensitivity (3.0 %, 95 %CI, 0.3-10.0 % in Jimma and 7.8 %, 95 %CI, 0.90, 21.8 % in Ambo abattoirs). Predilection site dissection has high specificity (98.9 %, CI, 96.3, 100.0), and Ag-ELISA has moderate specificity (86.9 %, CI, 77.6, 100). In this study, imperfect diagnostic tests were combined within a Bayesian framework to estimate true prevalence without relying on a perfect reference test. The findings highlight that the true prevalence of bovine cysticercosis is substantially higher than the apparent prevalence of the separate methods and that the sensitivity of current meat inspection practices is low. Therefore, adopting more sensitive diagnostics, routine incisions of the tongue and heart, and strengthened surveillance and training are suggested to improve detection and protect public health.