The purpose of the present study was to test the association between on-site intravenous fluid replacement and mortality in patients with severe trauma. The effect of prehospital time on this association was also evaluated. The design was that of an observational quasi-experimental study comparing 217 patients who had on-site intravenous fluid replacement (IV group) with an equal number of matched patients for whom this intervention was not performed (no-IV group). The patients were individually matched on their Prehospital Index obtained at the scene and were included in the study if they had an on-site Prehospital Index score > 3 and were transported alive to the hospital. The outcome measure of interest was mortality because of injury. The patients in the IV group had a significantly lower mean age (37 vs. 45 years; p < 0.001) and higher incidence of injuries to the head or neck (46 vs. 32%; p = 0.004), chest (34 vs. 17%; p < 0.001), and abdomen (28 vs. 12%; p < 0.001). The IV group also had a higher proportion of patients injured by motor vehicle crashes (41 vs. 27%; p = 0.003), firearms (9 vs. 2%; p = 0.001), and stabbing (20 vs. 9%; p = 0.001). The rate of extremity injuries (38 vs. 59%; p < 0.001) and falls (12 vs. 40%; p < 0.001) was lower for the IV group. In addition, the mean Injury Severity Score was significantly higher for the IV group (15 vs. 9; p < 0.001). The mortality rates for the IV and no-IV groups were 23 and 6% (p < 0.001). Logistic regression analysis showed that after adjusting for patient age, gender, Injury Severity Score, mechanism of injury, and prehospital time, the use of on-site intravenous fluid replacement was associated with a significant increase in the risk of mortality (adjusted odds ratio = 2.3; 95% confidence interval = 1.02-5.28; p = 0.04). To further evaluate the effect of prehospital time on the association between on-site IV use and mortality, the analysis was repeated separately for the following time strata: 0 to 30 minutes, 31 to 60 minutes, and >60 minutes. The adjusted odds ratios (95% confidence interval) for these strata were 1.05 (0.08-14.53; p = 0.97), 3.38 (0.84-13.62; p = 0.08), and 8.40 (1.27-54.69; p = 0.03). These results show that for prehospital times of less than 30 minutes, the use of on-site intravenous fluid replacement provides no benefit, and that for longer times, this intervention is associated with significant increases in the risk of mortality. The results of this observational study have shown that the use of on-site intravenous fluid replacement is associated with an increase in mortality risk and that this association is exacerbated by, but is not solely the result of, increased prehospital times. Our findings are consistent with the hypothesis that early intravenous fluid replacement is harmful because it disrupts the normal physiologic response to severe bleeding. Although this evidence is against the implementation of on-site intravenous fluid replacement for severely injured patients, further studies including randomized controlled trials are required to provide a definitive answer to this question.