This study addressed methodological issues of epidemiologic studies on venous thromboembolism (VTE) to show how design decisions can affect the results. It examines the influence of a restricting to hospitalized and to "idiopathic" cases of VTE in case-control studies on VTE risk and oral contraceptive (OC) use and thereby the bias potential. The diagnostic processes and OC exposures of 1068 suspected cases of VTE were followed in 21 German centers from 1994 to 1999 and evaluated compared to population and hospital controls. Of 606 confirmed cases, 385 (65.5%) were hospitalized, 168 (27.7%) were "idiopathic." Comparing users versus nonusers of OCs, the odds ratio (OR) for VTE risk for all cases and controls was 3.38 and 5.44 for "idiopathic" VTE. For hospital cases and controls, the OR was 3.72 and 9.1 for "idiopathic" VTE. The risk ratio for third- vs. second-generation OCs was increased in the hospital base but not in the population base. It was concluded that restriction to hospitalized events and exclusions of certain cases overestimate the VTE risk of OCs. An evidence-based consensus on methodological standards and definitions in case-control studies on VTE and steroid hormone use is required.