Background: A reduction in mortality with the early use of tranexamic acid has been demonstrated in severely injured patients who are bleeding. However, the modest treatment effect with no reduction in blood transfusion has raised concerns. The aim of the present study was to estimate the effectiveness of regular use of tranexamic acid in severely injured patients. Methods: This multicentre observational study used retrospectively collected data from consecutive injured patients (Injury Severity Score at least 16) treated in 15 Japanese academic institutions in 2012. A propensity score-matched analysis compared patients who did or did not receive tranexamic acid administration within 3 h of injury. Study outcomes included 28-day all-cause and cause-specific mortality, and need for blood transfusion. Results: Of 796 eligible subjects, 281 were treated with tranexamic acid. Propensity score matching selected a total of 500 matched subjects (250 in each group). Tranexamic acid administration was associated with lower 28-day mortality (10·0 versus 18·4 per cent; difference −8·4 (95 per cent c.i. –14·5 to −2·3) per cent) and lower 28-day mortality from primary brain injury (6·0 versus 13·2 per cent; difference −7·2 (−12·3 to −2·1) per cent). However, there was no significant difference between groups in the need for blood transfusion (33·2 versus 34·8 per cent; difference −1·6 (−9·9 to 6·7) per cent). Conclusion: Early tranexamic acid use was associated with reduced mortality in severely injured patients, in particular those with a primary brain injury.