Abstract: Evaluation plays an important role in security of cryptology, among which randomness is one of the most practical contents. There exist several test methods and software packages for randomness test now. But there isnt an integrated and applied quantitative evaluation model for manipulating the vast results at present. In this paper, randomness evaluation of cryptography is studied and block cipher is selected as a research instance. A tree-type index system for randomness is proposed by analyzing design principles of block cipher, and an evaluation model is built based on fuzzy multi-criteria decision-making. In this model, membership function is used to process randomness result, which can express the continuous and gradual character of randomness and can overcome the problem of information loss introduced by threshold method. This model has some advantages such as offering an effective method to quantitatively evaluate the randomness of block cipher, and providing a foundation of comprehensive evaluation of cryptography. The analysis also shows that the model is practical because its consumption of space and time is very low. Furthermore the model provides a general evaluation process for single index and attribute, and it can be easily modified to deal with the case of other fundamental types of cryptographic primitives, such as stream cipher.