Analytic hierarchy process (AHP) is put forward to make multiple attribute decision problem hierarchical. With the AHP, complicated decision system is structured, weightiness of the connection factor is compared, validity model is formed, and quantitative basis for decision and analysis is provided. The database preprocessing with AHP is to evaluate every database for data cleaning when the data warehousing is building. And the approach selects the appropriate database to preprocess so as to raise the efficiency of data preprocessing. Instead of the whole warehouse, using the more “valuable” databases for data mining based on the same theme gets similar effect and lower time complexity. So three data indexes, such as completeness, smoothness and consistency, are defined differently to incomplete, noisy, and inconsistent in data cleaning.