The details of the early star formation histories of tiny dwarf galaxies can shed light on the role in galaxy formation of the reionization which occured at high redshift. Isolated dwarfs are ideal probes since their evolution is not complicated by environmental effects owing to the vicinity of the Milky Way and M31. In addition, dwarf galaxies are the most common type of galaxies, and potentially the building blocks of larger galaxies. Since we can date the oldest stars in them, their study represents a complementary approach to the study of the formation and evolution of galaxies through high-z observations. We propose to use the ACS to obtain a homogeneus dataset of high-quality photometry, down to the old {13 Gyr} main-sequence turnoffs, for a representative sample of 4 isolated Local Group dwarf galaxies. These data are essential to unambiguously determine their early star formation histories, through comparison with synthetic color-magnitude diagrams, and using the constraints provided by their variable stars. Parallel WFPC2 observations of their halos will allow us to reveal the actual nature of their stellar population gradients, providing important aditional constraints on their evolution. The proposed observations are being complemented with ground-based spectroscopy, to obtain metallicity and kinematic information. The observations requested here, which must reach M_I=+3.5 {I=27.5-28.2} with S/N=10 in crowded systems, can only be achieved with HST using ACS, and won't be possible with planned ground- or space-based facilities such as JWST. Based on deep WFPC2 observations and ACS image simulations, our team has designed an observational strategy which carefully considers the optimal filter combination, the necessary photometry depth and the effects of stellar crowding.