We present constraints on the stellar-mass distribution of distant galaxies. These stellar-mass estimates derive from fitting population-synthesis models to the galaxies observed multi-band spectrophotometry. We discuss the complex uncertainties (both statistical and systematic) that are inherent to this method, and offer future prospects to improve the constraints. Typical uncertainties for galaxies at z ~ 2.5 are ~ 0.3 dex (statistical), and factors of ~ 3 (systematic). By applying this method to a catalog of NICMOS-selected galaxies in the Hubble Deep Field North, we generally find a lack of high-redshift galaxies (z > 2) with masses comparable to those of present-day ``L* galaxies. At z < 1.8, galaxies with L*-sized masses do emerge, but with a number density below that at the present epoch. Thus, it seems massive, present-day galaxies were not fully assembled by z ~ 2.5, and that further star formation and/or merging are required to assemble them from these high-redshift progenitors. Future progress on this subject will greatly benefit from upcoming surveys such as those planned with HST/ACS and SIRTF.