Abstract Biometric authentication plays an extremely important role all over the world over the last few decades. Biometrics literally means all technological techniques utilized to authenticate or identify persons relying on their physical and/ or behavioral characteristics. However, single biometric trait, such as face, iris, and fingerprint, usually fails to meet the security requirements of many applications with large population data. Therefore, multimodal biometrics, which employs two or more biometric modalities, has gained an increasing attention to overcome the problems associated with unimodal biometric system. This paper aims to propose a multimodal biometric authentication method based on the face and iris recognition First, the face and iris features are extracted separately using 2D wavelet transform and 2D Gabor filters, respectively. Second, the paper applies a feature-level fusion using a novel fusion method which employs both the canonical correlation process and the proposed serial concatenation. Finally, a deep belief network is used for the recognition process. The proposed system has been validated and its performance has been evaluated through a set of experiments on SDUMLA-HMT database. The proposed method has been compared with another method. The results of this paper have shown that the propped system has succeeded to achieve recognition accuracy up to 99%. It has shown a lower equal error rate (EER) and fusion time in comparison with other systems.