The virtual screening methods and techniques become one of the important and sophisticated ways of drug discovery, and molecules clustering, they are many methods proposed and applied in virtual screening, most of these screening methods used similarity coefficients to quantify the extent to which objects resemble one another. The result of using these similarity coefficients achieves a good result, and work is continuing to enhance and modify virtual screening methods or to present new methods, in this paper, we proposed a simple algorithm that uses simple equation that increased the effectiveness of virtual screening. The proposed method calculates the similarity and ranks the data to achieve better virtual screening results. We tested our proposed method with two benchmarks datasets Directory of Useful Decoys (DUD) data sets and Maximum Unbiased Validation (MUV) that already prepared and presented by 2D fingerprints ,the experiments have been conducted by selecting different 10 references from each activity class in each data sets , and we e evaluate the recall of active molecules at different at cut of 1% and 5% as usually done in virtual screening to evaluate the recall value, the overall results showed that the our proposed algorithm has good result in ligand-based virtual screening after comparing the result with Tanimoto coefficient which considered the standard similarity measure virtual screening.