Abstract: Open-pit coal mines are among the most drastic anthropogenic land disturbances. Using Wise County in the USA's Appalachian coal field as the study area, this paper evaluates the mine land disturbance and reclamation process over a 27-year period by conducting time-series analysis of multispectral remote-sensing data at the pixel scale. Twenty TM/ETM images obtained by the Landsat satellites, with 30 m spatial resolution, are treated as a multiple-year chronosequence. Polygonal vector files defining mining permitted areas and 6 high-resolution aerial images are used as auxiliary data; and the normalized difference vegetation index (NDVI) is used as a vegetative cover indicator. The methodology and study process include 3 steps. First, training data are prepared and multispectral image data are preprocessed. Data preprocessing includes band stacking, extracting the study area as image subsets, masking of cloud, cloud shadow and water, and computing NDVI based on each study-area pixel in each image. After the training data are generated, they are used to identify the NDVI thresholds for separating bare-ground from vegetated pixels, and ever-mined pixels from those un-mined. These separations are performed by visually inspecting each individual TM/ETM image displayed through a combination of the band of 2, 3 and 4 while referencing all available high-resolution aerial images and polygon vector files defining mining permitted area. Second, each pixel's time-series NDVI trajectory is constructed and analyzed. Based on individual inspection of training points' NDVI trajectory and qualitative trajectory classification, the characteristic parameters, including the NDVI maximum, the NDVI minimum, the difference of NDVI maxima before and after mining, the bare-ground threshold and ever-mined threshold, are designed, computed and used to identify whether the pixel is ever mined, revegetated or revegetated to the land cover level before mining. The relationship between NDVI minimal values and ever-mined thresholds for each study-area pixel over the full time series is computed and analyzed to determine if the pixel has been mined and, if so, the date of initial disturbance. For mined areas, the relationship between NDVI trajectory and bare-ground threshold for each point in time series after mining is analyzed to determinate if excavated land has been revegetated. For those revegetated pixels, the difference of NDVI maxima before and after mining is computed in order to find if it has been fully revegetated. Thirdly, by the data analysis function of database software and the spatial analysis function of ArcGIS software, the distribution of different mining-reclamation types is analyzed. Except for un-disturbed pixel, each mined pixel was classified for each year; such classes include the mined and un-vegetated during the whole observation period; the mined and un-revegetated after mining; the mined and revegetated without fully restored to the level before mining; and the mined and revegetated with vegetative cover fully restored to the level before mining. Time durations of non-vegetated status and revegetation after initial mining disturbance are also computed for fully reclaimed pixels. The conclusions are as follows: 1) During 1984-2010, the mining disturbs 45.80% mining permitted areas; 66.45% mined land is re-vegetated, which includes 38.42% mined land with fully restored vegetative cover and 28.03% mined land revegetated but without fully restored vegetative cover; 2) The fraction of mined lands with fully restored vegetative cover varies positively with mining-site age, which is found to be 85.54%, 13.92%, and 0.52% lands during the periods of 1984-1993, 1994-2003 and 2004-2010, respectively; 3) The average time for non-vegetated status and vegetated status prior to full cover restoration on the mined lands is 6 and 12 years, respectively; while the 25th, 50th, 75th and 90th percentiles for non-vegetated status are 4, 6, 10 and 16 years, respectively; 4) For full restoration of NDVI-derived vegetative cover to the level before mining, the time is 12 years on average. Study results demonstrate that the analysis of pixel-scale NDVI trajectories can reconstruct mining disturbance and land reclamation history, and can disclose its spatial heterogeneity over the broad expanse of space and time. Hence, such analyses can provide a scientific basis for land reclamation management and policy decisions. We note, however, that restoration of the NDVI to the level before mining is not a reliable indicator of reclamation success when viewed alone, although it is easily measured with satellite remote-sensing data. In order to interpret NDVI-derived vegetative cover data as indicators of reclamation success, other factors such as vegetation type and land productivity should be also considered.