Deep vein thrombosis (DVT) requires a prompt and accurate diagnosis to avoid the potentially fatal consequences of a delay in adequate treatment. The diagnostic strategy of suspected venous thromboembolism is an example of the application of Bayes' theorem. As a result, the available diagnostic tools, both noninvasive (such as ultrasonography, D-dimer, and computed tomography) and invasive (such as venography), are usually integrated in diagnostic strategies that require the evaluation of the a priori clinical probability of disease (pretest probability). These strategies also vary according to the patient characteristics that allow the definition of five different categories: symptomatic outpatients, symptomatic inpatients, high-risk asymptomatic subjects, patients with suspected recurrence, and pregnant women. Some areas are still a matter of debate, such as the clinical relevance of isolated distal DVT and the diagnosis of suspected ipsilateral recurrence.