Geostatistically based methods that utilize textural information are frequently used to analyze remote sensing (RS) images. The role of these methods in analyzing forested areas increased rapidly during the last several years following advancements in high-resolution RS technology. The results of numerous applications of geostatistical methods for processing RS forest images are encouraging. This paper summarizes such results. Three closely related topics are reviewed: 1) specific properties of geostatistical measures of spatial variability calculated from digital images of forested areas, 2) determination of biophysical forest parameters using semivariograms and characterization of forest ecosystem structure at the stand level, and 3) forest classification methods based on spatial information.