The significance of candiduria in critically ill patients remains unclear. It may represent harmless colonization or a potentially life-threatening infection. We analyzed 47 patients in the surgical intensive care unit (SICU) (trauma: 20, general surgery: 15, neurosurgery: 12) who had candiduria, defined by a colony count greater than 100,000/mL. Twenty-seven of these patients were studied retrospectively. Twenty were evaluated prospectively. All patients were receiving broad-spectrum antibiotics for bacterial infections. Retrospective group: ten patients (group A) did not develop disseminated candidiasis, whereas 17 patients (group B) did. Group B had higher APACHE II scores on admission (13.4 +/- 7.8) and at the time of candiduria (13.7 +/- 4.4) when compared with group A [admission: 5.0 +/- 4.6; candiduria: 6.7 +/- 3.6 (p < 0.02)]. In group B, disseminated candidiasis was not diagnosed and treated until 9.9 +/- 4.4 days after development of candiduria. Prospective group: twenty patients with candiduria were treated with systemic fluconazole (group C) at the time of candiduria. The APACHE II scores of group C on admission (12.8 +/- 3.9) and at the time of candiduria (10.5 +/- 4.0) were comparable with those of group B. No patient in Group C developed disseminated candidiasis. The septic mortality rates of groups A, B, and C were 0%, 53%, and 5%, respectively (p < 0.05-0.0001). In patients exhibiting ongoing sepsis and organ failure (high APACHE scores), candiduria may be an early indicator of systemic infection. Diagnosis of disseminated infection and its treatment may be delayed if conventional criteria for candidiasis (positive blood cultures, multiple site isolation) are awaited.(ABSTRACT TRUNCATED AT 250 WORDS)