Since late 1998, we have been making sustained measurements of the Earth's reflectance by observing the earthshine from Big Bear Solar Observatory. Further, we have simulated the Earth's reflectance for both the parts of the Earth in the earthshine and for the whole Earth. The simulations employ scene models of the Earth from the Earth Radiation Budget Experiment, simulated snow/ice cover, and near-real-time satellite cloud cover data. Broadly, the simulations and observations agree; however, there are important and significant differences, with the simulations showing more muted variations. During the rising phase of the Moon we measure the sunlit world to the west of California, and during the declining lunar phase we measure the sunlit world to the east. Somewhat surprisingly, the one third of the Earth to the west and that to the east have very similar reflectances, in spite of the fact that the topographies look quite different. The part to the west shows less stability, presumably because of the greater variability in the Asian cloud cover. We find that our precision, with steady observations since December 1998, is sufficient to detect a seasonal cycle. We have also determined the annual mean albedos both from our observations and from simulations. To determine a global albedo, we integrate over all lunar phases. Various methods are developed to perform this integration, and all give similar results. Despite sizable variation in the reflectance from night to night and from season to season (which arises from changing cloud cover), we use the earthshine to determine annual albedos to better than 1%. As such, these measurements are significant for measuring climate variation and are complementary to satellite determinations.