Bugs or even harmful behaviors may not be avoided in some software which is necessary to the user. While the proportion of the code space of such bugs and behaviors to the whole software is exceedingly low, the present malware analysis technology is not fit for the trusted control of such software. For this question, this paper presented a trusted control rule based on the action list as an idealized goal, then a second trusted control rule based on APIs was put forward to meet the engineering needs according to the certainty of workflow, the visibility of code and the computability of assets risk. And a trusted control model of subject's behaviors based on static measuring was setup with the latter rule to monitor the run of the user's necessary software trustily, which will prevent the bugs or even harmful behaviors ahead of schedule and ensure the software function just as the user will.