Software trustworthiness evaluation model is built upon the accurately eliciting of trustworthy requirements and the reasonable establish ment of indicator system in the domain special application. Toward software which has huge architecture and complex non-functional demands, trustworthy requirements become changing with software operational state transition. The stability of trustworthiness evaluation indicator system will be affected by trustworthy requirement dynamic evolution. The software trustworthiness evaluation and evolution problem are attracting wide attention in the field of trustworthy software. In this paper, a novel requirement-driven software trustworthiness evaluation and evolution model is designed. Firstly, several key technologies adopted in the process of software trustworthiness evaluation are analyzed and summarized, such as requirements analysis and indicators extraction, trustworthy evidence acquisition and conversion, etc. And the problem of trustworthiness evaluation adaptive solution under the requirements evolution is discussed. Secondly, incidence matrix is used to achieve correlation analysis between trustworthy attributes and then the variation rule of relative weight is revealed. On this basis, adaptive reconstruction device, which can analyze and solve the software trustworthiness evaluation indicator system of self-reconfiguration, is designed based on incidence matrix. Finally, a complete framework of trustworthiness evaluation evolution and evolution model is proposed. Experimental results show the rationality and validity of the model.