The antifungal susceptibilities of 598 isolates of Candida spp. (bloodstream and other sterile sites) to liposomal amphotericin B (L-AmB) versus amphotericin B (AmB) were determined. MICs were calculated using the Clinical and Laboratory Standards Institute broth microdilution (M27-A3) method for L-AmB and the Etest method for AmB. The MIC50/MIC90 (μg ml−1) values for L-AmB broth microdilution and AmB Etest were 0.25/1 and 0.19/0.5, respectively. The overall essential agreement (±2 dilutions) was 91.5 %, ranging from 37.5 % (Candida lusitaniae) to 100 % (Candida glabrata and Candida krusei). Categorical agreement between the two methods was categorized based on a previously published breakpoint (susceptible/resistant MIC cut-off of 1 μg ml−1). The overall categorical agreement at the 48 h reading was 97.3 %, ranging from 72.7 % (C. krusei) to 100 % (Candida albicans). Major and very major discrepancies occurred in 2.3 and 0.3 %, respectively. Spearman's ρ was 0.48 (P<0.0001). These results demonstrate the utility of the AmB Etest as a surrogate marker to predict the sensibility and resistance of Candida spp. to L-AmB and thus to support its use in antifungal treatment.