Abstract: In this paper, it proposed an underground air evaluation model based on the unascertained clustering. It referred to the underground air quality standards, and determined the pollution components, which included carbon dioxide, carbon monoxide, sulfur dioxide, nitrogen dioxide, and hydrogen sulfide. It used unascertained clustering to calculate the class center of each air rank, and determined the membership degree of the samples, then classified the samples. The evaluation of underground air quality was achieved. The results showed that the experimental data were consistent with the actual situation.