Abstract: This paper is based on lots of dynamic consolidation data in collapsible loess area. It uses the method of frequency statistics to determine factor weights according to the concept of variable weight. It sets up a case-based reasoning model in association with fuzzy analogy preferred ratio for predicting the effective depth of collapsible loess treated with dynamic consolidation. Examples of dynamic consolidation are taken as base cases, while examples that will be analyzed are taken as target cases in this model. Corresponding evaluation index is selected as the fuzzy factor. The similarity of fuzzy factors between base cases and target cases is calculated. The similar series are determined. The most similar base case to target case is found out in terms of the effective reinforced depth. Finally, the effective reinforced depth of dynamic consolidation can be predicted. It is shown from examples that the differences between the predicted and measured values are within 10%. Therefore, this is an effective method with high prediction precision and worth to be used.