Abstract Objectives Microbiological strategies are necessary to help clinicians discontinue empirical antifungal therapy in patients with suspected invasive candidiasis. Culture methods and biomarkers each show low sensitivity. We analysed the value of combining different biomarkers as a decision-making tool for discontinuing empirical antifungal treatment. Methods We studied stored serum samples from 31 patients with candidaemia (Candida albicans 40%, Candida tropicalis 20%, Candida parapsilosis 18%, Candida glabrata 12% and other 10%) and 50 patients with bacteraemia at Gregorio Marañón Hospital, Madrid, Spain. C. albicans germ tube antibody (CAGTA), mannan antigens (MN), antimannan antibodies (AMN) and (1→3)-β-d-glucan (BDG) were assayed using the manufacturer's and alternative cut-offs to improve the accuracy of the tests. Results The sensitivity of the biomarkers when used alone was low (58%−84%), but specificity was high (65.8%−92.0%). The best combinations were CAGTA and BDG using cut-offs of 1/80 and 80 pg/mL, respectively (sensitivity 96.8% and specificity 84%), and CAGTA and MN using cut-offs of 1/80 and 75 pg/mL, respectively (sensitivity 93.5% and specificity 86.0%). The sensitivity of both combinations was 100% for C. albicans, C. tropicalis and C. parapsilosis, but only combinations including BDG detected Candida krusei. The negative predictive values (NPVs) of both combinations were, respectively, 97.7% and 95.6% (prevalence of candidaemia, 23.6%). For a prevalence of candidaemia of 5% and 10%, the NPV reached 99.8% and 99.6%. Conclusions The combinations of CAGTA and BDG or CAGTA and MN had a very high NPV at the alternative cut-offs and could be used in antifungal stewardship programmes as a decision-making tool for discontinuing unnecessary empirical therapy in patients with suspected candidaemia.