Summary Background Aspirin is effective in the treatment of acute myocardial infarction and in the long-term prevention of serious vascular events in survivors of stroke and myocardial infarction. There is, however, no reliable evidence on the effectiveness of early aspirin use in acute ischaemic stroke. Methods The Chinese Acute Stroke Trial (CAST) was a large randomised, placebo-controlled trial of the effects in hospital of aspirin treatment (160 mg/day) started within 48 h of the onset of suspected acute ischaemic stroke and continued in hospital for up to 4 weeks. The primary endpoints were death from any cause during the 4-week treatment period and death or dependence at discharge, and the analyses were by intention to treat. 21 106 patients with acute ischaemic stroke were enrolled in 413 Chinese hospitals at a mean of 25 h after the onset of symptoms (10 554 aspirin, 10 552 placebo). 87% had a CT scan before randomisation. It was prospectively planned that the results would be analysed in parallel with those of the concurrent International Stroke Trial (IST) of 20 000 patients with acute stroke from other countries. Findings There was a significant 14% (SD 7) proportional reduction in mortality during the scheduled treatment period (343 [3·3%] deaths among aspirin-allocated patients vs 398 [3·9%] deaths among placebo-allocated patients; 2p=0·04). There were significantly fewer recurrent ischaemic strokes in the aspirin-allocated than in the placebo-allocated group (167 [1·6%] vs 215 [2·1%]; 2p=0·01) but slightly more haemorrhagic strokes (115 [1·1%] vs 93 [0·9%]; 2p>0·1). For the combined in-hospital endpoint of death or non-fatal stroke at 4 weeks, there was a 12% (6) proportional risk reduction with aspirin (545 [5·3%] vs 614 [5·9%]; 2p=0·03), an absolute difference of 6·8 (3·2) fewer cases per 1000. At discharge, 3153 (30·5%) aspirin-allocated patients and 3266 (31·6%) placebo-allocated patients were dead or dependent, corresponding to 11·4 (6·4) fewer per 1000 in favour of aspirin (2p=0·08). Interpretation There are two major trials of aspirin in acute ischaemic stroke. Taken together, CAST and the similarly large IST show reliably that aspirin started early in hospital produces a small but definite net benefit, with about 9 (SD 3) fewer deaths or non-fatal strokes per 1000 in the first few weeks (2p=0·001), and with 13 (5) fewer dead or dependent per 1000 after some weeks or months of followup (2p<0·01).