Using the analytic hierarchy process (AHP), we structured a hierarchical indicator system that brought about a set of grading criteria to evaluate forest quality of six forest types, including a monoculture Pinus tabulaeformis stand, a monoculture Quercus liaotungensis stand, a monoculture Betula platyphylla stand, a mixed stand of P. tabulaeformis and Q. liaotungensis, a mixed stand of P. tabulaeformis and B. platyphylla, and a mixed stand of Q. liaotungensis and B. platyphylla, in Caijiachuan Forest Station of Yanan, Shannxi Province, China. Results showed that this evaluation system was composed of three criteria and 8 indexes, which could be used to evaluate these secondary forests in this area. The three criteria were forest structure, stand productivity and regeneration status, respectively, among which forest structure attributes were the most important, followed by regeneration status, and then stand productivity. The relative contribution of each index to the forest quality estimation was in the order of canopy cover (0.356 2), litter depth (0.211 6), proportion of seedlings and saplings to the number of all dominant trees (0.130 5), Pielou index (0.105 8), stock volume (0.082 6), richness index (0.064 4), number of seedlings and saplings (0.037 2), and biomass of shrubs and grasses (0.011 8). This study also showed that the quality of most stands was more or less improved after the implementation of the natural forest protection project, but it was not consistent among forest types studied. More than half of the stands were classified as medium quality, and few stands were evaluated as poor quality. Three stand types, P. tabulaeformis, Q. liaotungensis and mixed P. tabulaeformis-Q.liaotungensis stands, had the highest synthesis evaluating value, with a range of 65-73, followed by the B. platyphylla stand, whereas the mixed stand of Q. liaotungensis-B. platyphylla had the lowest value (less than 60.00). The indicators and evaluation system developed in this study are easy to use for quantifying forest quality, and should be suitable for application in evaluating the secondary forests on Loess Plateau.