Forests are usually distributed on rugged mountains, therefore the elimination of topographic effects becomes one of the essential issues to improve monitoring accuracy of forest cover by remote sensing. Here we introduced a novel topography-adjusted vegetation index (TAVI) which is capable of removing topographic effects in rugged terrain based only on infrared and red wavebands data from optical remote sensing images, without support of digital elevation model (DEM) data. The Landsat TM images acquired in 1998 and 2008 respectively were utilized in case study to validate the TAVI performance and monitor the changes of forest cover with TAVI. From the regression analysis between TAVI and the solar incidence cosine and contrast analysis between TAVI and normalized difference vegetation index (NDVI) in the study area, it was discovered that TAVI can resist the topographic effects remarkably and was much better than NDVI in that the slope of linear regression equation of TAVI and the solar incidence cosine was only between ±0.04, and also their correlation coefficient was just between ±0.08. The forest cover images in rugged areas computed from TAVI displayed a planar spatial distribution instead of texture pattern resulting from topography. The dynamic monitoring results turn out that forests in the research area grow in a generally sustainable way during 1998--2008.