Field studies of meat inspection procedures in three different slaughter houses were performed in Germany, in 1996. In total 22,634 fattening pigs from 63 farms were inspected twice, using a visual and the traditional (mandatory inspection system) procedure. Statistical analysis was performed using the rate of ND+ (Non‐Detected Positives) for both visual and traditional inspection. Neither the visual nor the traditional methods detected all lesions which were supposed to be in the lot of animals inspected. Some of the results could be explained by the nature and obviousness of the respective lesion. Conclusions regarding the future performance of inspection should not be drawn before a formal risk analysis procedure for the particular lesion has been done.