Histone deacetylases (HDACs) have emerged as important anti-tumor targets in recent years. As HDACs comprise multiple isoforms and there are different physiological functions among various isoforms, the development of selective HDAC inhibitors has attracted a great deal of attention. This study focused on the discovery of selective HDAC1, HDAC8 inhibitors and specifically a homology model for HDAC1 was generated. A comparison was made between the HDAC1 homology model and the crystal structure of HDAC8, which showed that some active site residues were different from each other and these residues played important roles in the selectivity between HDAC1 and HDAC8. Two linear regression models were established based on the inhibitory activities against HDAC1, HDAC8 and the docking scores of 52 compounds. The developed linear regression models for HDAC1 and HDAC8 have non-cross validated correlation coefficients R2 of 0.82 and 0.80, respectively, which indicates that the results are statistically significant. These models were used to predict the activities of the synthesized compounds and these prediction results can provide further insights into the selectivity of HDAC1, HDAC8.