The aim of this study was to analyze the behavior of landscape metrics for detecting changes in forests of Himalayan foothills. Satellite data of 1990, 2001 and 2006 with 30 m resolution have been used to prepare forest cover (3 classes) and land use (7 classes) map using supervised classification technique. The landscape metrics used to quantify spatial patterns of the changes has been calculated using Fragstats 3.3. The number of landscape metrics available for such studies are very large. Knowledge of interrelationship and redundancy between the metrics are required for the right selection and interpretation of the landscape change. A set of landscape metrics were calculated that provide overall perspective of the class level changes in the landscape. Multivariate statistical techniques have been used to identify independent components and to cluster them. 3D pattern space was formed by three different suitable metrics for different classes. From the result, it is found that landscape indicators help to successfully explain the dynamic landscapes. There was strong indication of deforestation. Settlement increased very rapidly but in a planned way.