OBJECTIVE: The clinical characteristics and prognosis of patients treated for Candida peritonitis (CP) were compared according to the type of systemic antifungal therapy (SAT), empiric (EAF) or targeted (TAF) therapies, and the final diagnosis of infection. METHODS: Patients in intensive care units (ICU) treated for CP were selected among the AmarCAND2 cohort, to compare patients receiving EAF for unconfirmed suspicion of CP (EAF/nonCP), to those with suspected secondarily confirmed CP (EAF/CP), or with primarily proven CP receiving TAF. RESULTS: In all, 279 patients were evaluated (43.4% EAF/nonCP, 29.7% EAF/CP and 25.8% TAF patients). At SAT initiation, the severity of illness was similar among EAF/nonCP and EAF/CP patients, lower among TAF patients (median Simplified Acute Physiology Score II (SAPS II) 49 and 51 versus 35, respectively; p 0.001). Candida albicans was involved in 67%, Candida glabrata in 15.6%. All strains were susceptible to echinocandin; 84% to fluconazole. Echinocandin was administered to 51.2% EAF/nonCP, 49% EAF/CP and 40% TAF patients. At day 28, 72%, 76% and 75% of EAF/nonCP, EAF/CP and TAF patients, respectively, were alive. An increased mortality was observed in patients with a Sequential Organ Failure Assessment (SOFA) score <7 if SAT was delayed by ≥6 days (p 0.04). Healthcare-associated CP (OR 3.82, 95% CI 1.52-9.64, p 0.004), SOFA ≥8 at ICU admission (OR 2.61, 95% CI 1.08-6.34; p 0.03), and SAPS II ≥45 at SAT initiation (OR 5.08, 95% CI 1.04-12.67; p 0.001) impacted the 28-day mortality. CONCLUSIONS: In summary, only 56.6% of ICU patients receiving SAT had CP. Most strains were susceptible to SAT. A similar 28-day mortality rate was observed among groups; the late administration of SAT significantly worsened the prognosis of patients with less severe CP.