To determine whether clinical features can be used in a decision rule to prospectively identify a subgroup of internal jugular catheter placements that are correctly positioned and free from mechanical complications, thus obviating the need for routine postprocedural chest radiographs in selected patients. Prospective cohort study. Tertiary care teaching hospital. A total of 107 consecutive patients who presented to our catheter service for internal jugular catheter insertion because of clinical indications between November 1995 and April 1996. Exclusion criteria were mechanical ventilation, an altered mental status, an age of <15 years, and a height of <152 cm. Right or left internal jugular vein catheter placement followed by a postprocedural chest radiograph. The operating physician completed a detailed questionnaire for each catheter insertion, designed to detect potential complications and to predict the necessity, or lack of necessity, for a postprocedural chest radiograph. The questionnaire documented patient characteristics, the number of needle passes, difficulty establishing access, operator experience, poor anatomical landmarks, number of previous catheter placements, resistance to wire or catheter advancement, resistance to aspiration of blood or flushing of the catheter ports, sensations in the ear, chest, or arm, and development of signs or symptoms suggestive of pneumothorax. After catheter insertion, chest radiographs were obtained to assess for mechanical complications and malpositioned catheters. In 46 cases, the decision rule predicted either a complication or a malposition and, thus, the need for a chest radiograph. In 61 cases, neither was predicted (no chest radiograph was needed). Radiographs confirmed one complication (pneumothorax) and 15 catheter tip malpositions (nine in the right atrium and six in the right axillary vein). Among the 46 cases predicted to have a potential complication or malposition, there were one actual complication (pneumothorax) and six actual malpositions (three axillary vein malpositions and three right atrial malpositions). The positive predictive value of this decision rule is 15%. Among the 61 cases predicted to be free from complications or malpositions and not to require a postprocedural chest radiograph, there were nine unexpected malpositions (three axillary vein malpositions and six right atrial malpositions). The negative predictive value is 85%. The overall sensitivity of the decision rule for detecting complications and malpositions is 44%, and the specificity is 55%. In experienced hands, internal jugular venous catheterization is a safe procedure. However, the incidence of axillary vein or right atrial catheter malposition is 14%, and clinical factors alone will not reliably identify malpositioned catheters. Chest radiographs are necessary to ensure correct internal jugular catheter position.