While deep vein thrombosis (DVT) may be clinically suspected at several time points after major orthopedic surgery, clinical examination is often unreliable, and compression ultrasonography (CUS) screening at discharge is of limited value. A prospective cohort study was carried out in 1,033 consecutive patients who had undergone major hip or knee surgery, aimed to assess the accuracy of a strategy consisting of clinical investigation followed by CUS in the detection of proximal DVT before discharge. The circumferences of both legs were measured in all patients; those exhibiting >2 cm difference between them were considered to have suspected DVT, and underwent bilateral CUS. The same diagnostic workup was repeated on days 45 and 90 after surgery. Three patients developed pulmonary embolism (PE) during admission (one died). Five additional patients died before discharge. Routine clinical evaluation before discharge was done in 1,025 patients, and 105 (10%) had suspected DVT. CUS confirmed the diagnosis in 24 (2.3% of the overall series). After discharge, 59 patients had suspected DVT on day 45, 53 on day 90. DVT diagnosis was confirmed by CUS in 27 (26%). Three additional patients developed PE (1 fatal). This translates into a sensitivity of the routine examinations at discharge of 44%. A limited diagnostic workup for DVT before discharge has the capacity to identify 44% of those patients who will become symptomatic afterwards.