Study objectives: To determine the prevalence of endogenous and exogenous risk factors for venous thrombosis in patients with upper limb deep vein thrombosis (DVT), and to evaluate the risk of clinically detectable pulmonary embolus, recurrent DVT, and postphlebitic symptoms in these patients. Design: A combined prospective and retrospective descriptive analysis of a cohort of patients with upper limb DVT compared with age- and sex-matched patients with lower limb DVT. Setting: Internal medicine departments, and hematology and vascular surgery outpatient clinics at a tertiary-care university hospital. Patients: Consecutive patients with 'spontaneous' upper limb DVT diagnosed between 1989 and 1997 were studied. Twenty age- and sex-matched patients with lower limb DVT admitted to the hospital via the emergency department served as control patients. Results: Eighteen patients with upper limb DVT were studied. An endogenous risk factor (thrombophilia) was present in 11 of 18 patients vs 8 of 20 control patients (p = not significant). In the upper limb group, nine patients had activated protein C resistance, four patients had anticardiolipin antibodies, and two patients had both forms of thrombophilia. Furthermore, 14 of the upper limb DVT patients were found to have an exogenous risk factor for thrombosis compared with 7 of the patients with lower limb DVT (p = 0.01), and 66.6% of patients with upper limb DVT had both an exogenous and an endogenous risk factor for thrombosis vs 15% of patients with lower limb DVT (p < 0.002). No clinically detectable pulmonary emboli occurred among the upper limb DVT patients. Three patients have minor postphlebitic symptoms. Two patients experienced recurrent DVT. Conclusion: In the majority of patients with upper limb DVT that we studied in this relatively small study, exogenous (environmental) or endogenous risk factors for venous thrombosis, or a combination of both, were found. Furthermore, in our patients, these thromboses had a low propensity to cause clinically significant pulmonary embolus and did not cause significant postphlebitic symptoms. Finally, we suggest that anticoagulant therapy for these thromboses may be adequate and that thrombolytic agents and surgical intervention are not routinely indicated.