A randomised controlled trial was conducted to assess whether bladder catheters with preconnected sealed junctions were associated with a lower risk of urinary-tract infection than were catheters without such junctions, and to determine whether prevention of catheter-associated infection would be accompanied by a reduction of mortality. Among those not taking systemic antibiotics, patients assigned sealed junction catheters had fewer infections and deaths. Before they received antibiotics, the risk of infection among those assigned unsealed catheters was 2.7 times that of patients assigned sealed catheters (95% confidence interval=1.3-5 . 4, p=0 . 007). Among the 220 patients who received no antibiotics, 14% (15/108) of those assigned unsealed catheters and 4% (4/112) of those assigned sealed catheters died. Stratification by important risk factors for mortality yielded an adjusted risk ratio for death of 3.4 (95% CI=1.1-10.7, p=0.03). Among patients who received systemic antibiotic the use of sealed catheters did not affect infection rates (RR=0.9, 95% CI=0.5-1.5, p=0.68) or deaths (RR=1.2, 95% CI=0.6-2.2, p=0.62). These data indicate how the rates of infection and mortality can be reduced in hospital. Since the degree of reduction in mortality corresponded with the degree of reduction of infection, measures to prevent catheter-associated nosocomial urinary tract infection should be implemented.