Gas inflows and outflows are central to cosmological models of galaxy formation and evolution. Inflow has long been recognized as a requirement for explaining the stellar metallicity distribution and the star formation timescale in late-type galaxies. More recently, tuning the feedback prescription has emerged as the critical factor for modeling the galaxy mass function and gas-phase, mass-metallicity relation; the results of these numerical simulations have provided much insight into which galaxies might lose significant amounts of mass and metals via galactic winds. The challenge now is to test these theoretical solutions against empirical measurements. The talk will take a detailed look at galactic gas flows in major mergers, an example of the brief evolutionary phase when bulges and supermassive black holes grow rapidly. I will argue that the fastest outflowing gas is not seen in the circumnuclear region but is found condensing out of a fast, hot wind. Over the era of disk assembly, our lookback studies demonstrate that gas outflow and inflow were common among all star- forming galaxies, not just the starbursts. These gas flows are invisible in emission at large radii, but I will show how we are studying their gas kinematics in absorption using new spectroscopy of background quasars. The combined results support the mass- loading factors that simulations require for galaxies similar in mass to the Milky Way, draw attention to some properties of gas flows that may be missing from models, and suggest further comparisions between observations and simulations.