Candidaemia is associated with poor outcomes including high mortality rates. Controversy remains regarding whether fluconazole or an echinocandin is the optimal choice for initial candidaemia treatment, particularly among high-risk patients such as the immunocompromised or critically ill. To understand optimal initial treatment of candidaemia. We conducted a retrospective study of immunocompromised or ICU adult patients with candidaemia from 2010 to 2014. Patients who received ≥3 consecutive days of initial treatment with fluconazole or micafungin were included. The primary outcome was complete response at day 14, defined as clinical improvement and blood culture sterilization. Secondary outcomes included microbiological and clinical success, survival and recurrent candidaemia. A total of 197 patients were included; 76 received fluconazole and 121 received micafungin. There was no difference in complete response between the fluconazole and micafungin groups (ICU: 38% versus 40%, P = 0.87; immunocompromised: 57% versus 59%, P = 0.80). Secondary outcomes including survival were also similar. In multivariable analysis, among ICU patients, Pitt bacteraemia score < 4 (P = 0.002) and time to antifungal (P = 0.037) were associated with meeting the primary outcome; white blood cell count > 11 cells × 103/µL on day 0 (P < 0.001) and Candida isolated from a non-blood site (P = 0.025) were associated with not meeting the primary outcome. Among immunocompromised patients, white blood cells > 11 × 103/µL (P = 0.003) and Candida isolated from a non-blood site (P = 0.026) were associated with not meeting the primary outcome. These data suggest that among ICU or immunocompromised patients, severity of illness rather than initial antifungal choice drove clinical outcomes.