Satellite remotely sensed observations provide a unique capacity to monitor biospheric activity at regional and global scales. Spectral vegetation indices from these observations capture the heterogeneity of growth patterns across the Earth's surface including the significant decline in vegetation activity above 50 degrees N latitude. Conversion of these observations to meaningful assessments of vegetation growth, particularly changes within and between years, requires more than a simple linear transform between the spectral vegetation index and forest production. Although satellite observations of vegetation indices are essential, appropriate environmental variables and models are also necessary. The approach is illustrated by a study of the southern Boreal forest in northern Minnesota. A conceptual and methodological framework for monitoring primary production from remotely sensed observations alone is proposed and results of its application in the Global Production Efficiency Model (GLO-PEM) for global boreal forests are discussed.