Detection of the fungal cell wall component beta-glucan (BG) in serum is increasingly used to diagnose invasive fungal infections (IFI), but its optimal use in hematology patients with high risk of IFI is not well defined. We retrospectively analyzed the diagnostic accuracy, optimal cut-off level, and potential confounding factors of BG reactivity. The inclusion criteria were: adult patients with hematologic disease who were admitted to the hematology ward during the 2-year study period and who had two or more consecutive BG assays performed. In total, 127 patients were enrolled. Thirteen patients with proven or probable IFI, as defined by the 2008 European Organization for Research and Treatment of Cancer/Mycoses Study Group (EORTC/MSG) criteria, were identified. Receiver operating characteristic (ROC) curve analysis showed a high overall diagnostic performance (area under the ROC curve = 0.98) and suggested an optimal cut-off level of 158 pg/ml, with a sensitivity and a specificity of 92 % and 96 %, respectively. Multiway analysis of variance indicated that treatment with pegylated asparaginase (p < 0.001), admission to the intensive care unit (ICU; p = 0.0007), and treatment with albumin, plasma, or coagulation factors (p = 0.01) are potential confounding factors of BG reactivity. We propose that a higher cut-off level than that recommended by the manufacturer should be used to monitor adult hematology patients at high risk for IFI. Our results also suggest that elevated BG levels in patients treated with pegylated asparaginase, albumin, plasma, or coagulation factors, or those admitted to the ICU should be interpreted with caution.