Cell penetrating peptides (CPPs) are kinds of small molecular peptides, which can penetrate cell membrane easily. Their discovery sheds light on delivering drugs into target efficiently. Therefore, the investigation of CPPs has crucial significance in biomedicine field. In this work we intended to find factors which have influence on CPPs penetrating activity and introduce a new method called physico-chemical mass center of bio-sequence. This method was enlightened by particle mechanics. We used it to analyze bio-sequence. In addition, the differences between all kinds of CPPs with different penetrating activities were also investigated. The CPPs and Non-cell penetrating peptides (NonCPPs) were obtained based on CPPsite database version 1.0 (http://crdd.osdd.net/raghava/cppsite1/) and references, respectively. After that we extracted CPPs with high, medium, and low penetrating activities (HCPPs, MCPPs, LCPPs), respectively. Firstly, the amino acids composition was calculated for every peptide in HCPPs, MCPPs, LCPPs datasets, individually. We used ANOVA to analyze the amino acids composition in HCPPs, MCPPs, LCPPs datasets. From the results we noticed that there has a significant difference for amino acids with electric charge and hydrophobic properties. These results revealed that electrostatic and hydrophobic interactions may play a key role in CPPs' penetrating activities. ANOVA analysis also revealed that helix and coil structure may also have influence on CPPs' activities. Secondly, CPPs and NonCPPs can yield a clustering phenomenon under some physicochemical features such as pK values, and their corresponding lengths. HCPPs, MCPPs, LCPPs, and NonCPPs can be divided into three groups. The differences between CPPs and NonCPPs can be reflected by the above phenomena. Lastly, as a case study we used the new concept named physico-chemical mass center of bio-sequence method encouraged by particle mechanics to study CPPs. The sequences can be represented by a series of points in the above method. Combining with principal component analysis (PCA), we found that the distribution of CPPs under their first, second and third scores can generate a cluster for CPPs. The differences between HCPPs, MCPPs, LCPPs and NonCPPs were also demonstrated by the above phenomenon. From this study, we can conclude that this work can help us to further understand the differences between them in terms of their primary structures and the factors affecting on their activities. Most important, the physico-chemical mass center can be used to analyze other biological issues as well as input features for biological pattern recognitions.