One hundred and fifteen patients were entered into a prospective randomized trial to establish whether midline abdominal incisions through the umbilicus produced any differences in the incidence of wound infection or failure when compared with the more conventional midline incision skirting around the umbilicus. No significant difference was shown between wound infection rates in 58 transumbilical and 51 circumumbilical incisions. Follow-up of surviving patients for at least one year showed no difference in incisional hernia rate between the two groups of patients. No problems were encountered in making the incisions through the umbilicus, but skirting the umbilicus was felt to cause problems with slewing of the scalpel blade, and asymmetry of the scar in 13 circumumbilical incisions. This study has shown that the practice of avoiding the umbilicus in midline abdominal surgery serves no useful purpose.