Invasive candidiasis, including candidemia and deep-seated Candida infections, is a severe opportunistic infection with an overall mortality in ICU patients comparable to that of severe sepsis/septic shock. With an incidence ranging from 5 to 10 cases per 1000 ICU admissions, invasive candidiasis represents 5–10% of all ICU-acquired infections. Although a high proportion of critically ill patients is colonised with Candida spp., only 5–40% develop an invasive infection. The occurrence of this complication is difficult to predict and an early diagnosis remains a major challenge. Indeed, blood cultures are positive in a minority of cases and often late in the course of infection. New non-culture based laboratory techniques may contribute to early diagnosis and management of invasive candidiasis. Recent data suggest that prediction rules based on risk factors, clinical and microbiological parameters or monitoring of Candida colonisation may efficiently identify critically ill patients at high risk of invasive candidiasis who may benefit of preventive or pre-emptive antifungal therapy. In many cancer centres, exposure to azoles antifungals has been associated with an epidemiological shift from Candida albicans to non-albicans Candida species with reduced antifungal susceptibility or intrinsic resistance. This trend has not been observed in recent surveys on candidemia in non-immunocompromised ICU patients. Prophylaxis, pre-emptive or empirical antifungal treatment are possible approaches for prevention or early management of invasive candidiasis. However, the selection of high-risk patients remains critical for an efficient management aimed at reducing the number needed to treat and thus avoiding unnecessary treatments associated with the emergence of resistance, drug toxicity and costs.