The high rates of invasive candidiasis among intensive care unit (ICU) patients suggest that antifungal prophylaxis might be of value, but rules identifying patients who would best benefit are not established. Based on a retrospective study of 327 patients who stayed in a surgical ICU for > or = 4 days and had an 11.0% rate of invasive candidiasis, we sought to identify useful predictive rules. As prior work suggests that prompt initiation of prophylaxis is of value, we required our rules to be based on patient data routinely available during the week prior to ICU admission through the third day of the ICU stay. Patients with any combination of diabetes mellitus, new onset hemodialysis, use of total parenteral nutrition, or receipt of broad-spectrum antibiotics had an invasive candidiasis rate of 16.6% versus a 5.1% rate for patients lacking these characteristics (P = 0.001). Fifty-two percent of patients staying > or = 4 days in the ICU met this rule and the rule captured 78% of the patients who eventually developed invasive candidiasis. Risk-stratified antifungal prophylaxis in the ICU is possible. Validation of these results in other types of ICU is now needed.