The learning curve characteristics of lead extraction with a laser sheath were examined in 19 patients. Forty-two leads were removed: five leads were removed by traction alone, five required a femoral approach as a primary procedure and a laser sheath was used in 32 lead extraction attempts. Primary laser success was achieved in 26 attempts (81%). A femoral approach was successfully applied as a back-up procedure in five of the failures. Overall, 26% of the leads were removed by the femoral approach. The overall success rate was 98% (41 of 42 leads). No variables related to the patients, leads, or extraction techniques were significantly related to failure of laser sheath extraction. There was a distinct learning curve with all but one failure occurring in the first half of our cases. All failures occurred with leads implanted from the right subclavian vein. In four, a sharply angled curve at the subclavian vein-superior vena cava junction could not be passed with the laser sheath. The ability to smooth this curve improved the results during the learning curve. All procedures were performed in the operating room for safety reasons. This precaution was lifesaving in a case of acute tamponade after laser extraction of an atrial lead. In another case the left internal mammary artery was torn after laser sheath extraction, causing the formation of a false aneurysm. New pacing leads were introduced in nine patients during the same procedure. The mean procedure time was 255 +/- 110 min. reflecting the complexity of these procedures.