Venous thromboembolism (VTE) is extraordinarily common and is a major cause of morbidity and mortality. However, accurate and timely diagnosis of VTE is confounded by its kaleidoscopic presentation. Clinical prediction rules (CPRs) and D-dimer testing have both been increasingly employed to clarify the complex decision making required in such cases. Formal clinical pretest probability now serves as the root of algorithms for the diagnosis of DVT and PE. A low pretest probability of VTE plus a negative D dimer can be combined in a bayesian fashion to effectively exclude the diagnosis of VTE. The evidence for this strategy is strongest in younger outpatients with no associated comorbidities, no prior history of VTE, and a short duration of symptoms.