Slope is the most important topographic variables. It is a critical work to establish a scientific and reasonable slope classification systems for terrain research and slope mapping. After a comprehensive summary to the previous works in this field, it is pointed out that slope classification scheme could be divided into three different types on the basis of their serving purpose, namely critical classification, equal interval classification and fuzzy classification. Their availability is analyzed after explaining the characteristics of the slope classifications. Taking Jiuyuangou area as the test area representing the loess hill gully terrain type, a digital slope model (DSM) was extracted from 1∶10 000 scaled DEMs. Different slope spectrums can be derived from DSMs according to the different slope classification systems, which shows a interesting discrepancy both in their distribution patterns and statistic values.