We present a comprehensive approach for deriving the star formation history of the universe self-consistently with a wide array of observations (stellar mass functions, stellar mass clustering, specific star formation rates, and the cosmic star formation rate) and with merger rates from N-body simulations in the context of LCDM cosmology. Our approach explores a wide parameter space of systematic uncertainties, allowing a broader range of possible star formation scenarios at high redshifts; these uncertainties include aspects (e.g., a steeper faint-end slope for the stellar mass function) which help resolve tensions between the cosmic star formation rate and stellar mass functions for LBGs at z>1. We present derived constraints on the star formation rates and histories for galaxies as a function of stellar and halo mass from z=0 to z=8. We show that constraints from observations favor a clear change in the star formation rates of massive galaxies around z=2, consistent with a transition from cold-mode to hot-mode accretion at that redshift and very old stellar populations at the present day. We find, on the other hand, that low-mass galaxies do not have a similar break and that many have star formation histories which have been increasing ever since the galaxies first formed. We also discuss the largest uncertainties on current constraints and the direction that future surveys should take to best increase our understanding. In summary, we provide for the first time a complete, consistent picture of the evolutionary path of galaxies over 96% of the history of the universe, with profound implications for galaxy simulations, semi-analytic models, and our understanding of star formation in the cosmos.