Abstract: The Yellow River Delta is a typical eco-environmental fragile region where soil quality is not high and salinization is universal, and it is of great significance for the rehabilitation and reconstruction of degraded soil to know the situation of degradation soil quickly and accurately in the region. This research chose Kenli County in the Yellow River Delta as the research area and the measured data in 2008 as the basis, and the comprehensive evaluation of soil degradation was carried out based on soil degradation composite index, which was built by weighted combination of soil degradation evaluation index weight and membership value. Using TM image that was at the same time as the measured data, the spectral characteristics of different soil degradation levels were researched, and the correlations between soil degradation composite index and the bands' gray values were analyzed; with the combination of qualitative and quantitative methods, the sensitive bands were screened. Then the sensitive spectral indices of soil degradation were built, based on which the remote sensing inversion model of soil degradation was established, and the exponential model was selected ultimately which had the highest fitting precision. Verified on the accuracy and the inversion results, the model was applied to the remote sensing images in 2011 and 2013, and the soil degradation dynamics were analyzed during 2008-2013 at last. The results showed the lower-degree degradation soil in the study area was far away from the ocean, and degradation degree of soil in close distance from ocean was relatively high; taken altogether, the extent of soil degradation from the coast to the inland in the study area showed a trend from high to low. The sensitive bands were TM1, TM2 and TM3, and the spectral indices of soil degradation based on the combination of these 3 sensitive bands were characteristic spectral indices of soil degradation. The regression function models were built by soil degradation spectral indices and soil degradation composite index, and the exponential model was the best, which had the highest F value and the best fitting. The remote sensing inversion model of exponential form based on the characteristic spectral indices of soil degradation had more higher accuracy, whose R2 was 0.7182, and validation RMSE, relative error and determination coefficient were 0.0241, 3.66% and 0.6724, respectively. By contrast, the inversion results were consistent with the comprehensive evaluation results based on the measured data in the same year; in particular, the inversion results were more ideal in the area where the land types were relatively simple and the spectral information was relatively clear. The condition of soil degradation changed little in the study area during 2008-2013, and the specific performances were that the area of mild degradation soil increased and degradation soil at other grades decreased, a small amount of moderate degradation soil turned into mild, and there was a gradual improvement in the trend as a whole. This study provides technical support for monitoring the degradation soil in the Yellow River Delta, and provides decision-making basis for the sustainable utilization and protection of land resources in this area.