Historically patients with severely depressed or elevated white blood cell (WBC) counts during infection were felt to have worse outcomes. To test this assumption we prospectively analyzed all infections on the surgical services at the University of Virginia hospital between December 1, 1996 and April 1, 1999. Among 1737 infectious episodes 59 presented with leukopenia (WBC count < or = 3,000 cells/microL) whereas 66 presented with leukemoid responses (WBC count > or = 30,000 cells/microL). Compared with other infected patients leukopenic patients had higher Acute Physiology and Chronic Health Evaluation II (APACHE II) scores (18+/-0.9 vs 12+/-0.2, P < 0.0001) and mortality (23.7% vs 11.4%, P = 0.004). Patients with leukemoid responses also had higher APACHE II scores (21+/-1.0 vs 12+/-0.2, P < 0.0001) and mortality (30.3% vs 11.4%, P < 0.0001). Compared with a control group randomly matched (2:1) by age and APACHE II score, however, there was no significant difference in mortality associated with leukopenia or a leukemoid response. Furthermore logistic regression did not reveal leukopenia or leukemoid responses to be independent predictors of mortality (odds ratio for death with leukopenia = 1.57, 95% confidence interval = 0.63-3.91, P = 0.33; odds ratio for death with leukemoid response = 1.19, 95% confidence interval = 0.70-2.02, P = 0.53). Although very low or very high WBC counts may represent markers of severe illness in infected surgical patients they do not appear to be significant contributors to a worsened outcome.