The pharmacodynamic and pharmacokinetic (PK-PD) properties of amphotericin B (AmB) formulations against invasive pulmonary aspergillosis (IPA) are not well understood. We used an in vitro model of IPA to further elucidate the PK-PD of amphotericin B deoxycholate (DAmB), liposomal amphotericin B (LAmB) and amphotericin B lipid complex (ABLC). The pharmacokinetics of these formulations for endovascular fluid, endothelial cells, and alveolar cells were estimated. Pharmacodynamic relationships were defined by measuring concentrations of galactomannan in endovascular and alveolar compartments. Confocal microscopy was used to visualize fungal biomass. A mathematical model was used to calculate the area under the concentration-time curve (AUC) in each compartment and estimate the extent of drug penetration. The interaction of LAmB with host cells and hyphae was visualized using sulforhodamine B-labeled liposomes. The MICs for the pure compound and the three formulations were comparable (0.125 to 0.25 mg/liter). For all formulations, concentrations of AmB progressively declined in the endovascular fluid as the drug distributed into the cellular bilayer. Depending on the formulation, the AUCs for AmB were 10 to 300 times higher within the cells than within endovascular fluid. The concentrations producing a 50% maximal effect (EC50) in the endovascular compartment were 0.12, 1.03, and 4.41 mg/liter for DAmB, LAmB, and ABLC, respectively, whereas, the EC50 in the alveolar compartment were 0.17, 7.76, and 39.34 mg/liter, respectively. Confocal microscopy suggested that liposomes interacted directly with hyphae and host cells. The PK-PD relationships of the three most widely used formulations of AmB differ markedly within an in vitro lung model of IPA.