We studied the influence of the lipid/amphotericin B (AMB) ratio and the phospholipid type on the in vitro renal cell toxicity and antifungal efficacy of lipid-associated AMB (L-AMB). L-AMB was prepared at one of two different lipid/AMB ratios (1 and 40) by incubating AMB with empty small unilamellar vesicles, made from one of three different phospholipids: dipalmitoyl-, dimirystoyl-, and distearoylphosphatidylcholine (DPPC, DMPC, and DSPC, respectively). Renal cell toxicity, investigated through an assessment of the Na-dependent uptake of phosphate by proximal tubular cells, and fungicidal effect against Candida albicans were studied after 1 h of treatment at 37 degrees C. The amount of unbound AMB present in each L-AMB formulation was studied by use of circular dichroism. At a lipid/AMB ratio of 40, the three lipidic formulations were not toxic for renal cells but were less effective against C. albicans than AMB; however, DSPC-AMB, which contained 50% unbound AMB, was more effective against C. albicans than DPCC-AMB or DMPC-AMB, containing 0 and 13% unbound AMB, respectively. At a lipid/AMB ratio of 1, the antifungal effects of L-AMB and AMB were similar, whatever the phospholipid used, but only DMPC-AMB remained highly protective against AMB renal cell toxicity, despite the presence of the same amount of unbound AMB (50%) in DMPC-AMB and DPPC-AMB. We conclude that the in vitro activities and renal cell toxicities of different L-AMB formulations are influenced by the phospholipid type and the lipid/AMB ratio. The optimal ratio depends on the phospholipid itself. At a lipid/AMB ratio of 40, the antifungal activity depends mainly on the amount of unbound AMB in the formulation. At a lipid/AMB ratio of 1, the renal cell toxicity also depends on the fluidity of the phospholipid.