System runtime environment and multiple external factors together with internal multientity information flow mutual interference can break system credibility, and result in unexpected outputs. Existing research mainly aims at the integrity measurement of entities under the initialized trusted hardware environment, failing to consider the trusted influence brought by the confidentiality, and the frequency of the trusted measurement of entities cannot be synchronized with the progress. We propose a multilevel dynamic trusted measurement model based on information flow theory. By using the basic idea of intransitive noninterference theory of information flow as reference and introducing a trusted proxy module, we design a multilevel security access control policy, hence the trusted measurement of entities can be measured dynamically from aspects of entity integrity and confidentiality. We describe the formal description and trusted proof of the model and verify the model through an abstract system example. Compared with existing research, it has a better realtime measurement performance, and it is a contextaware fine-grain trusted measurement model.