In this study, two models for fingerprint and DNA identifications are constructed based on modern technologies, while offering significant advances over prior models. Our models have high credibility, obtaining relatively accurate results under different circumstances. Under different assumptions, this model tests the probability of the validity in the statement that human fingerprints are unique to be 93.94%. In other words, the percentage of misidentification is 6.06%. This model is a robust fingerprint identification method that can tolerate highly nonlinear deformations. The model is tested on the basis of a self-built database, proving that the model has high credibility, and convincing results are obtained from sensitivity analysis. In order to estimate the odds of misidentification by DNA evidence, we emphasized on two factors that might contribute to misidentification: random match probability and the probability arising from laboratory errors. Then, a model is developed using Bayes’ theorem to reveal the inherent relationship between them, which carries some reference value. The probability of matching by DNA evidence is estimated based on the changes in the significant level. Finally, the probabilities of misidentification by both fingerprint evidence and DNA evidence are compared using numerous data. We found that the probability of the former is 6.06% and that of the latter is smaller than 4.0 x 10?10. Therefore, it can be concluded that DNA identification is far better than that of fingerprint identification.