In this paper, we considered the problem of sample imbalance in the context of gene selection. Based on simple random sampling, two evaluation models were proposed to investigate the effect of sample imbalance on gene selection. Under the proposed evaluation models, the performances of five famous gene selection methods on the unbalanced data were compared. The experimental results indicated that the proposed evaluation models are effective and the sample imbalance has a great influence on gene selection. Our findings provide some guidelines in the design of microarray experiments and the following data analysis, and two evaluation models are suitable for selecting feasible gene selection method to identify differential expression genes.