Abstract: Multibiometric systems are expected to be more accurate due to the presence of multiple evidences. Score level fusion is the most commonly used approach in multibiometrics. There are usually three kinds of techniques of score fusion: transformation-based, classifier-based and density-based. This paper firstly analyzes the advantages and disadvantages of the three types of algorithms. Then a novel confidence-based fusion technique is proposed which combines the advantages of transformation-based and density-based fusion strategies. The confidence is based on the false reject rate (FRR) and false accept rate (FAR) which can cultivate the distribution of the match scores while avoiding computing the posterior probability. The comparison between the new algorithm and the existing representative algorithms is conducted in experiments. The experimental results show that the new fusion scheme is robust for different multibiometric systems.