Different therapeutic strategies for invasive fungal diseases have been explored, each with particular strengths and weaknesses. Broad-spectrum antifungal prophylaxis seems logical, but selective use is important due to its substantial disadvantages, including interference with diagnostic assays, selection for resistance, drug toxicity and drug-drug interactions. Antimould prophylaxis should be restricted to high-risk groups, such as patients undergoing intensive chemotherapy for acute myeloid leukaemia or myelodysplastic syndrome, allogeneic HSCT patients with prior invasive fungal infection, graft-versus-host-disease or extended neutropenia, recipients of a solid organ transplant, or patients with a high-risk inherited immunodeficiency. An empirical approach, whereby mould-active therapy is started in neutropenic patients with fever unresponsive to broad-spectrum antibiotics, is widely applied but incurs the clinical and cost penalties associated with overtreatment. A benefit for all-cause mortality using empirical therapy has not been shown, but it is recommended for high-risk patients who remain febrile after 4-7 days of broad-spectrum antibiotics and in whom extended neutropenia is anticipated. There is growing interest in delaying antifungal treatment until an invasive fungal infection is confirmed ('pre-emptive' or 'diagnostics-driven' management), prompted by the development of more sensitive diagnostic techniques. Comparisons of empirical versus pre-emptive regimens are sparse, particularly with modern triazole agents, but treatment costs are lower with pre-emptive therapy and the available evidence has not indicated reduced efficacy. Pre-emptive treatment may be appropriate in neutropenic patients who remain febrile after administration of broad-spectrum antibiotics but who are clinically stable. Further work is required to define accurately the specific patient subgroups in which each management approach is optimal.