To study the discharge performance of Li-O 2 batteries, we propose a multiscale modeling framework that links models in an upscaling fashion from nanoscale to mesoscale and finally to device scale. We have effectively reconstructed the microstructures of a Li-O 2 air electrode in silico, conserving the porosity, surface-to-volume ratio, and pore size distribution of the real air electrode structure. The mechanism of rate-dependent morphology of Li 2 O 2 growth is incorporated into the mesoscale model. The correlation between the activesurface-to-volume ratio and averaged Li 2 O 2 concentration is derived to link different scales. The proposed approach's accuracy is first demonstrated by comparing the predicted discharge curves of Li-O 2 batteries with experimental results at the high current density. Next, the validated modeling approach effectively captures the significant improvement in discharge capacity due to the formation of Li 2 O 2 particles. Finally, it predicts the discharge capacities of Li-O 2 batteries with different air electrode microstructure designs and operating conditions.