The estimation of the hemispherical reflectance and the instantaneous albedo of canopies from top of canopy satellite reflectance data was investigated. The study was designed to approximate the specifications of generic sensors aboard satellites like NOAA, VEGETATION, MERIS, MISR, MODIS, and PRISM. These sensors acquire reflectance data in two to six wave bands distributed along the visible, near-infrared, and middle infrared domains. Five great biomes (grassland, sparse vegetation, tropical forest, boreal forest, and bare soil) were approximated, simulating the corresponding top of canopy reflectances as observed from the satellites using well-known leaf, soil, and canopy radiative transfer models, including the effect of cloud cover that limits the actual data acquisition scheme. Albedo was accurately derived from the hemispherical reflectance observed in only a few wave bands. When using six wave bands, albedo was estimated within 1% relative accuracy. The MRPV bidirectional reflectance distribution function (BRDF) model was tested to derive the hemispherical reflectance from the top of canopy bidirectional data as sampled by the generic sensors during a 32 day orbit cycle. Results showed that this is the main source of error, with a relative accuracy around 6%. This showed the importance of the directional sampling scheme and possible improvements that may be made to the model and the way it is fitted to the observed data. The algorithm developed produced a relative accuracy less than 7% for the albedo estimation, when using the six wave bands and a ¿45¿ across-track directional scanning capacity. The results were discussed with particular emphasis on the optimization of sensors and algorithms dedicated to albedo estimation and to the use of hemispherical reflectance as a potential normalized geophysical product for monitoring vegetation. ¿ 1999 American Geophysical Union