Abstract: The traditional DEM accuracy evaluation models have some defects in scientificity and applicability. For instance, the wisely used elevation mean error model only shows the overall differences in elevation in the given area, but ignores the fact that DEM representative error still exists even when there is no difference in elevation between a demo DEM and an reference DEM. This research puts forward a novel DEM accuracy evaluation model based on cut fill method in GIS. The DEM cut fill error is defined as the volumetric differences between the DEM and reference DEM divided by the area of the DEM region. Based on the concept and calculation method of DEM cut fill error, we then explored the relationship between E_c and DEM resolution R and the average terrain slope S, in order to find out how DEM resolution and DEM average slope affect the DEM accuracy. Experiments are carried out based on reference DEM with one meter resolution in different landscapes including plains, hills and mountains. We calculate the DEM cut fill error of every DEM part referencing to accurate DEM with our proposed calculation method. Together with the resolution and average terrain slope of every DEM part, quantitative calculation model is found to calculate the DEM cut fill error, which is expressed as E_c=0.004 8·R·S. Finally, further verification experimental results show that the estimation accuracy to predict DEM cut fill error of the quantitative model is more than 95.85%. This model is useful in deciding proper DEM resolution to meet the tolerances of error under different terrain conditions.