To investigate the natural history of unruptured aneurysms and predictive risk factors determining subsequent rupture, the authors followed 142 patients with 181 unruptured aneurysms until death or subarachnoid hemorrhage intervened, or for at least 10 years after the unruptured aneurysm was diagnosed. Six patients had a symptomatic aneurysm, five had an incidentally discovered aneurysm, and 131 had multiple aneurysms, of which the ruptured lesion was clipped at the beginning of the follow-up study. The median follow-up time was 13.9 years (range 0.8 to 30.0 years). During 1944 patient-years of follow-up study there were 27 first episodes of hemorrhage from a previously unruptured aneurysm, giving an average annual rupture incidence of 1.4%. Fourteen of these bleeding episodes were fatal. The cumulative rate of bleeding was 10% at 10 years, 26% at 20 years, and 32% at 30 years after the diagnosis. The only predictor for the rupture was the size of the aneurysm (p = 0.036). However, in patients with multiple aneurysms (the main subgroup) the only variable that tended to predict rupture was the age of the patient: risk of rupture was inversely associated with age (p = 0.080). The median diameter of the aneurysms was 4 mm at the beginning of the follow-up period, both in those with and those without a later hemorrhage. During the angiographic monitoring period, a ruptured aneurysm significantly (p < 0.001) increased in size in 17 patients with hemorrhage but aneurysms did not increase significantly in 14 patients without hemorrhage. In addition, a new aneurysm was found in six of 31 patients. The authors conclude that an unruptured aneurysm should be operated on, irrespective of its size, if it is technically possible and the patient's age and concurrent diseases are not contraindications to surgery.