Diagnosis of bloodstream infections in critically ill patients is difficult. This case control study involved a total of 22 patients with confirmed bloodstream infections and 44 concurrent controls from an intensive care unit in Western Australia. We aimed to assess whether eosinopenia and C-reactive protein are useful markers of bloodstream infections in critically ill patients. The patients with bloodstream infections had a more severe disease and a longer length of intensive care unit (10.7 vs 4.0 days, P = 0.001) and hospital stay (40.9 vs 17.9 days, P = 0.015) than the controls. Univariate analyses showed that C-reactive protein (area under the receiver operating characteristic curve 0.847, 95% confidence interval (CI) 0.721 to 0.973), eosinophil counts (area under the receiver operating characteristic curve 0.849, 95% CI 0.738 to 0.961) and fibrinogen concentrations (area under the receiver operating characteristic curve 0.730, 95% CI 0.578 to 0.882) were significant markers of bloodstream infections. C-reactive protein concentration was, however the only significant predictor in the multivariate analysis (odds ratio 1.21 per 10 mg/l increment, 95% CI 1.01 to 1.39, P = 0.007). C-reactive protein concentration appears to be a better marker of bloodstream infections than eosinopenia in critically ill patients. A large prospective cohort study is needed to assess whether eosinopenia is useful in addition to C-reactive protein concentrations as a marker of bloodstream infections.