We identified 33 patients with definite pacemaker endocarditis--that is, with direct evidence of infective endocarditis, based on surgery or autopsy histologic findings of or bacteriologic findings (Gram stain or culture) of valvular vegetation or electrode-tip wire vegetation. Most of the patients (75%) were > or = 60 years of age (mean 66 +/- 3; range 21 to 86). Pouch hematoma or inflammation was common (58%), but other predisposing factors for endocarditis were rare. At the time that pacemaker endocarditis was found, the mean number of leads was 2.4 +/- 1.1 (range 1 to 7). The interval from the last procedure to diagnosis of endocarditis was 20 +/- 4 months (range 1 to 72). Endocarditis appeared after pacemaker implantation, early (< 3 months) in 10 patients and late (> or = 3 months) in 23 patients. Fever was the most common symptom, being isolated in 36%, associated with a poor general condition in 24%, and associated with septic shock in 9%. Transthoracic echocardiography showed vegetations in only 2 of 9 patients. Transesophageal echocardiography demonstrated the presence of lead vegetations (n = 20) or tricuspid vegetations (n = 3) in 23 of 24 patients (96%; p <0.0001 compared with transthoracic echocardiography). Pulmonary scintigraphy showed a typical pulmonary embolization in 7 of 17 patients (41%). Pathogens were mainly isolated from blood (82%) and lead (91%) cultures. The major pathogens causing pacemaker endocarditis were Staphylococcus epidermidis (n = 17) and S. aureus (n = 7). S. epidermidis was found more often in early than in late endocarditis (90% vs 50%; p = 0.05). All patients were treated with prolonged antibiotic regimens before and after electrode removal. Electrode removal was achieved by surgery (n = 29) or traction (n = 4). Associated procedures were performed in 9 patients. After the intensive care period, only 17 patients needed a new permanent pacemaker. Overall mortality was 24% after a mean follow-up period of 22 +/- 4 months (range 1 to 88). Eight patients who were significantly older (74 +/- 3 vs 63 +/- 3 years; p = 0.05) died < or = 2 months after electrode removal, whereas 25 were alive and asymptomatic.