Six hundred and four patients with atherothrombotic cerebral ischemic events (transient: 16% or completed: 84%) referable either to the carotid or to the vertebral-basilar circulation were entered into a double blind randomized clinical trial (AICLA) to determine whether aspirin (A) (1 g/day) or aspirin (1 g) + dipyridamole (225 mg) (AD) would produce a significant reduction in the subsequent (3 years) occurrence of fatal and non fatal cerebral infarction. Randomization produced remarkably comparable treatment groups and this good comparability was maintained throughout the study. Adherence to the protocol and drug compliance were excellent. Side effects, particularly peptic ulcers and bleedings of various origin, were significantly (p less than 0.03) more frequent in the two treatment groups containing aspirin. At the end of the study (3 years), the number of fatal and non fatal cerebral infarction was 31 in the P group (placebo), 17 in the A group and 18 in the AD group. Taking into account the duration of follow up for each patient, these figures correspond to cumulate rates of 18% in the P group and 10.5% in the 2 others. Analysis with the Mantel Method showed: a difference at the 6% level between the 3 groups and between P and AD; a difference at the 5% level between P and A; no difference between A and AD; a difference at the 2% level between the P group and the two treated groups taken together (A + AD).(ABSTRACT TRUNCATED AT 250 WORDS)