The diagnosis of venous thromboembolism (VTE) is complex, and many cases of pulmonary embolism (PE) and deep vein thrombosis (DVT) go undetected despite validated diagnostic algorithms. This study evaluated the diagnostic performance of compression ultrasound (CUS) when systematically performed in patients admitted to an internal medicine department for dyspnea and/or respiratory failure. We conducted a prospective observational cohort study of consecutive adult hospitalized patients admitted for dyspnea and/or respiratory failure with at least one of the following: tachycardia (> 100 bpm), tachypnea (> 20/min), chest pain, cough, syncope, or hemoptysis. Patients with a previous diagnosis of VTE or who underwent computed tomography pulmonary angiography (CTPA) or CUS during evaluation in the emergency department were excluded. The study included 263 patients (50.2% women, average age 84 years). CUS was positive in 31 patients (11.8%); Bilateral DVT was diagnosed in two patients and unilateral DVT in 29 patients. Of these, 10 underwent CT scan, with PE confirmed in 9 cases. Using the Wells score for DVT (cut-off ≥ 2), only 8 patients (25.8%) were at high risk. The accuracy of the Wells score in identifying PE was suboptimal, as 5 of 9 patients (55.5%) with confirmed PE were in the low-risk group (three-level interpretation) and 8 (89.9%) were in the “EP unlikely” group (two-level interpretation). The systematic use of CUS as a point-of-care tool can improve the diagnostic accuracy for VTE in patients admitted to internal medicine departments with dyspnea/respiratory failure.