The enormous case-fatality rate of severe sepsis and septic shock has resulted in considerable efforts being made towards understanding their complex mechanisms of pathogenesis. This has been done with the hope that agents that interfere with the pathways of pathogenesis and modulate the immune response of the host may be candidates for therapy. Disappointing results from most trials of immunomodulators in sepsis have led to understanding that the progression of patients to multiple organ dysfunction syndrome involves blunting of the pro-inflammatory cytokine storm. Instead, the compensatory anti-inflammatory response syndrome (CARS) develops, which is characterised by immunoparalysis. Components of this syndrome are impaired phagocytosis by neutrophils, decreased expression of HLA-DR on monocytes, impairment of ex vivo cytokine stimulation of monocytes, CD4 lymphopenia due to apoptosis of lymphocytes and predominance of anti-inflammatory T h2 and regulatory T-cell responses over pro-inflammatory T h1 and T 17 responses. CARS is not the sole explanation for the failure of trials of immunomodulators in sepsis. Recent data from the Hellenic Sepsis Study Group demonstrate that components of CARS upon transition from sepsis to severe sepsis/shock differ in relation to the underlying type of infection. These data underscore that the pathogenesis of sepsis presents considerable heterogeneity from one patient to another. That heterogeneity should be taken into consideration when deciding to administer an immunomodulator.