Between 1988 and 2000, 378 cases of keloids were treated in the authors' department, and 147 keloids in 129 patients were selected for this study. Keloids that occurred at a different site in the same patient and keloids that recurred later at the same site were deemed to be different keloids. Those keloids were surgically removed, and the patients were treated postoperatively with 15-Gy electronbeam irradiation and followed for more than 18 months. The therapeutic outcomes were evaluated. Statistical analysis was performed using Fisher's exact probability test or chi-square test. Recurrence occurred in two sites on 14 earlobes (14.3 percent), in two sites on 12 necks (16.7 percent), in 22 sites on 51 anterior chest walls (43.1 percent), in 13 sites in 33 scapular regions (39.4 percent), in four sites on 15 upper limbs (26.7 percent), in four sites in 11 suprapubic regions (36.4 percent), and in one site on 1 1 lower limbs (9.1 percent). The overall recurrence rate was 32.7 percent. Analysis of the therapeutic outcomes showed that the recurrence rates in the sites with high stretch tension, such as the chest wall, and the scapular and suprapubic regions were statistically higher than in sites without high tension, such as the neck, earlobes, and lower limbs (41.1 percent versus 13.5 percent, p = 0.0017). The results suggested that keloid sites with a high risk of recurrence should be treated with escalated radiation doses and posttreatment self-management. (Plast. Reconstr. Surg. 111: 547, 2003.)