This paper presents a novel chaotic encryption technique for the secure transmission of biometrics data. The proposed technique relies on the fractional random wavelet transform (FrRnWT), the modular transform and the generalized logistic map (GLM). The core idea is to decompose a biometric image into a fractional random wavelet domain using the chaotically generated transform orders. The FrRnWT coefficients are then diffused by a chaotically generated matrix key. In the penultimate step, the modular transform is applied on the diffused image to get the encrypted biometrics image. A reliable decryption process is proposed to reconstruct the original biometrics from the encrypted image. The experimental results and analysis demonstrate the efficiency and robustness of the proposed technique. © 2014 Elsevier B.V.