Many linear or non-linear statistics models have been developed for the estimation of fractional vegetation coverage by using vegetation indices. However, as the disturbance by uncertainty factors such as various crop planting density and nitrogen application, vegetation indices are limited to monitor regional vegetation coverage. In this paper, vegetation indices inversion models of fraction vegetation coverage based on regression analysis method were established and evaluated by using observed hyperspectral reflectance and vegetation coverage data set of winter wheat in the year 2010—2011. Firstly, the empirical models’ applicability (sensitivity, interannual stability and accuracy) were analyzed by using noise equivalent and model evaluation parameters. Simulation results indicated that there is a better result of using a second order polynomial regression equation to describe relationships between vegetation indices NDVI (Normalized difference vegetation index), TSAVI (Transformed soil adjusted vegetation index) and fraction vegetation coverage. While vegetation indices MSAVI (Modified soil adjustment vegetation index) and EVI (Enhanced vegetation index) exhibited a linear relationship with various fraction vegetation coverage. Evaluation results showed that: the correlation coefficient of regressed evaluation equations between predicted and measured vegetation coverage (Fc) were a little lower than the former modeling equations. All the evaluation relationships were significant at p=001 confidence level, which indicated these vegetation indices inversion models seemed stable among years and could give simple but reliable estimate of fraction vegetation coverage in this region. Sensitivity analysis suggested that under low to medium coverage (0~60% Fc) conditions, if the local soil information was available, using TSAVI to estimate variation of vegetation coverage showed better performance. However, if there was no information on soil characteristics, NDVI could assure estimation accuracy of fraction vegetation coverage. When vegetation cover Fc>60%, MSAVI was suggested to be used for estimating vegetation coverage, which displayed better sensitivity, stability and accuracy. Then, the general linear model (GLM) was employed to analyze the residuals of empirical models under conditions of various planting densities and nitrogen application rates. The results were somewhat inspiring: under condition of adequate water supply, all four vegetation indices (NDVI, EVI, TSAVI, MSAVI) exhibited no sensitive to various planting densities and nitrogen application rates during the entire growth period of winter wheat. This means models based on these four vegetation indices may not require re parameterization when apply to crops with different planting densities and nitrogen application rates. The regional winter wheat coverage could be directly estimated by using vegetation indices inversion models under the circumstances of abundant water supply. These findings provide a theoretical and technical support for the use of vegetation index to quickly and accurately estimate the regional vegetation coverage. However, as the regional land surface could be various and changeable, this paper could only explain the strength of vegetation indices inversion models for adequate water supply conditions, further studies are required for assessing vegetation indices method applicability in different crop intercropped and water and fertilizer coupling conditions.