Abstract: Index weight is a key factor in the evaluation of forest health. Exploring a scientific method for determining the index weight provides a new approach to evaluate forest health and a theoretical basis to improve the recognition of the evaluation of forest health. Here, we developed a structural equation model for assessing forest health of three typical forest types (Chinese fir pure forest, Chinese fir mixed forest, and natural secondary forest), in which forest structure, forest productivity, soil condition, and stand stability were set up as the firstorder factors and forest health as the second order factor. The index weight in the structural equation model was determined based on the conduction mechanism. The structural equation model was used to assess the health status of the three forest types. The results showed that: (1) For the structural equation model, the chi square degree of freedom (χ²/df) was 1.793, ranging 1-3; the root mean square error of approximation (RMSEA) was 0.079; the comparative fit index (CFI), normed fit index (NFI) and TackerLewis index (TLI) were 0.943, 0.908 and 0.929, respectively. The modelling results were compatible with the observed data. The structural equation model ran well and satisfied research requirements. (2) The index weights in the forest health evaluation system were 0.2515, 0.2263, 0.2277, and 0.2944 for forest productivity, forest structure, stand stability, and soil condition, respectively, indicating that soil condition was the key factor driving forest health. (3) The comprehensive scores of forest health for the examined three forest types ranged 2.1772-4.1708, indicating that the health status of the three forests were in the health, subhealth and medium-health status. In general, the health degree of the three forests ranked as natural secondary forest > Chinese fir mixed forest > Chinese fir pure forest. Our findings suggest that the structure equation model had advantages of extracting all information from original data and overcoming the multicollinearity uncertainties. The structural equation model provided a scientific basis and reasonable approach for the assessment of forest health. Results of health assessment showed that the three forest health evaluation indices could objectively reflect forest health status.

Key words: stand, forest health, index weight, structural equation model.