We propose a method of extracting cryptographic key from dynamic handwritten signatures that does not require storage of the biometric template or any statistical information that could be used to reconstruct the biometric data. Also, the keys produced are not permanently linked to the biometric hence, allowing them to be replaced in the event of key compromise. This is achieved by incorporating randomness which provides high-entropy to the naturally low-entropy biometric key using iterative inner-product method as in Goh-Ngo, and modified multiple-bit discretization that deters guessing from key statistics. Our proposed methodology follows the design principles of block ciphers to result in unpredictable key space and secure construction.