Mortality significantly increases in patients with candidaemia who receive inappropriate fluconazole therapy. The goals of this study were to compare hospital length of stay and costs for non-neutropenic patients with candidaemia treated with fluconazole based on the empirical dose and time until initiation of therapy. A retrospective cohort study was conducted of patients with candidaemia who were prescribed fluconazole at the onset of candidaemia or later. Hospital-related costs were compared based on time to initiation of fluconazole therapy and empirical fluconazole dose. A total of 192 non-neutropenic patients (55% male; mean age+/-standard deviation, 56+/-17 years) were identified. Isolated Candida species included C. albicans (59%), C. glabrata (15%), C. parapsilosis (11%), C. tropicalis (6%), C. krusei (3%) or other Candida spp. (6%). Time to initiation of fluconazole was Day 0 (35.4%), Day 1 (14.1%), Day 2 (26.6%) or Day >or=3 (23.9%). Thirty-two patients (17%) received a dose of fluconazole >or=6 mg/kg on Day 0. Total costs were lowest for patients started on fluconazole on the culture day with adequate doses ($35,459+/-25,988) compared with all other patients ($52,158+/-53,492) (P=0.0088). After controlling for covariates, each 1-day delay in fluconazole therapy was associated with increased total hospital costs of $6392+/-3000 (P=0.0344), and an adequate fluconazole dose was associated with decreased total hospital costs of $18,744+/-7173 (P=0.0097). A delay or an inadequate dose or fluconazole in patients with candidaemia was associated with increased hospital costs. Improved methods to diagnose patients with candidaemia quickly are needed.