Despite multiple trials of potential therapies for sepsis, no strategies have yet been persistently shown to have beneficial effects on outcomes. The main reason for the disappointing results is that patient populations in these studies have been too heterogeneous. Selecting patients on the basis of general symptoms is not enough. Rather patients should be selected according to the likely action of the drug in question. To achieve this, improved biomarkers of sepsis and of the immune response are needed and the activities of the individual agents need to be carefully characterized. New candidates are being developed and the results of ongoing and recent clinical trials of immunomodulatory therapies are eagerly awaited as new therapies for sepsis are urgently needed.