A population of 199 patients from Rochester, MN, was followed from the time of their first carotid or vertebral-basilar transient ischemic attack (TIA). Patients treated with anticoagulants had no significant difference in survival from untreated patients. Among patients with carotid TIA who received anticoagulants, the net probability of stroke was slightly but not significantly lower than in untreated patients. The difference favoring treated patients with vertebral-basilar TIA was significant starting at three months. The rate of intracranial hemorrhage was higher higher among all patients receiving anticoagulant treatment than among untreated patients and was significantly higher among those 55 to 74 years old. Almost all the hemorrhages occurred after a year or more of anticoagulant treatment and in patients more than 65 years old. Patients with high diastolic blood pressure had a significantly higher net probability of stroke than did patients with lower blood pressure and those receiving antihypertensive drugs. By implication, treatment of hypertension was effective in preventing stroke in patients with TIA. Linear discriminant analysis and actuarial analysis indicated that diastolic blood pressure and anticoagulant therapy were the only factors that influenced stroke occurrence. There was no suggestion that previous myocardial infarction, angina pectoris, valvular heart disease, cardiac arrhythmia, or congestive heart failure--individually or in combination--influenced the occurrence of stroke or survival.