Fish samples were collected from 33 sampling sites (8 reference sites and 25 observed sites) in Liao River Basin, China, during August, 2009. Twenty three candidate metrics including species richness, Shannon-Wiener diversity index, % of water column species, % of Cobitidae species, % of Cypriniformes species, % of Perciformes species, % of Cyprinidae species, % of benthic individuals, number of total individuals, % of omnivorous individuals, % of insectivorous individuals, % of phytophagous individuals, % of carnivorous individuals, % of tolerant individuals, %of intolerant individuals, % of viscid eggs spawning species, % of drift eggs spawning species, % of epipelagic eggs spawning species, % of demersal eggs spawning species, % of special spawning mode species, % of nesting and guarding eggs species, % abnormal individuals, were evaluated stepwise by distributing range analysis, discriminatory power analysis and Pearson′s correlation analysis. Basing these analysis we obtained 9 metrics to produce IBI evaluation. The traditional step method and ration scoring method were used separately to calculate the value of each metric and IBI. Based on the 25% percentile of IBI value in reference sites, the criteria of health ranking was determined. The distribution range below 25% percentile was divided into 3 quarters, forming 4 grades: health, fair, poor and very poor. In the follow, Pearson′s correlation analysis was used to reveal the correlation between IBI scores and environment factors. Results were shown as follows: Ten metrics had the highest discriminatory power between reference sites and observed sites, which were species richness, Shannon-Wiener diversity index, % of water column species, % of Cobitidae species, % of benthic individuals, number of total individuals, % of omnivorous individuals, % of tolerant individuals, %of intolerant individuals, % of viscid ovum individuals. Pearson’s correlation analysis was performed on all data from 33 sites for the above ten metrics. If metrics that measured the same criterion were correlated significantly (|R|>0.9), they should not be used together to determine impairment. Therefore, the % of Cobitidae species index was rejected because of its distinct correlation with % of benthic individuals index, considering the widely use of the latter in most references. The results of the traditional step method showed 3 healthy sites, 10 fair sites, 15 poor sites and 4 very poor sites, while, ration scoring method showed 6 healthy sites, 6 fair sites, 14 poor sites and 6 very poor sites. Though the results of the two methods were not exactly the same, they were not in contradiction, and the ultimate trend was consistent. They showed the condition of the water in most part of Liao River Basin was poor or very poor. The Pearson′s correlation analysis showed that the IBI was significantly positive correlated with the elevation, habitat value, and significantly negative correlated with CODCr, chloride, total dissolved granule, sand contents, hardness. The results indicate that that ecosystem health of Liao River Basin was strongly related to chemical pollution, water loss and soil erosion and habitat destruction, which should be paid more attention.