We investigated the role of therapeutic dose monitoring (TDM) in the treatment of fungal infections with voriconazole through 49 analyses of 34 patients who received treatment for hematologic diseases. Voriconazole concentration was highly variable among patients regardless of renal, liver functions, or age, and the effect of dose enhancement was not constant. This indicates the difficulty of predicting voriconazole concentration without TDM. We evaluated the outcome with the composite assessment system where patients were assumed non-responders when they failed to show improvement in at least 2 of the following 3 criteria: clinical, radiologic, and mycologic. We showed that concentration-response relationship depended on the status of underlying hematologic diseases; this relationship was observed only in cases without refractory hematologic diseases, but not in those with refractory diseases. In the former group, cases with >2 mg/L of concentration were associated with good response to voriconazole. On the other hand, elevation of hepatic enzyme was frequently observed when voriconazole concentration was >6 mg/L. From these results, we concluded that TDM should be executed and targeted to 2-6 mg/L to improve efficacy and to avoid side effects.