The presence and severity of lung lesions recorded post-mortem is commonly used as an indicator to assess the prevalence of respiratory problems in batches of bovines. In the context of a welfare monitoring based on on-farm measures, the recording of clinical signs on calves at the farm would be more convenient than the recording of lung lesions at slaughter. The aim of the present study was to investigate the relationship between clinical respiratory signs at farm and post-mortem analyses of lung lesions observed at slaughter in veal calves. If clinical signs were a good predictor of lung lesions it could be possible to integrate only those measures in a welfare monitoring system. One-hundred-and-seventy-four batches of calves were observed 3 times: at 3 and 13 weeks after arrival of the calves at the unit and at 2 weeks before slaughter. For each batch a maximum of 300 calves was observed and the proportions of calves showing abnormal breathing, nasal discharge and coughing were recorded. Post-mortem inspection was carried out on a sample of lungs belonging to calves from the observed batches. Each examined lung was classified according to a 4-point scale for pneumonia from healthy lung (score 0) to severe lesions (score 3). The clinical signs recorded infra vitam were significantly correlated with moderate and severe lung lesions for observations at 13 weeks and 2 weeks before slaughter and the level of the correlation was highly variable (r(sp) from 0.16 to 0.40). Receiver operating characteristic (ROC) curves were created and the area under the curves showed that batches with a high proportion of lungs with moderate or severe lesions could not be accurately detected by the three clinical signs of respiratory disorders. These results suggest that both clinical signs and post-mortem inspection of lung lesions must be included in a welfare monitoring schemes for veal calves.