Abstract: The monitoring and controlling of workflow process is an efficient method for dealing with the exception of workflow in run time and guaranteeing the enaction of workflow instances in the way of correctness and effectiveness. The monitoring and controlling of workflow based on states not only can cope with the exceptions from progression of workflow instances at real time, eliminate the retardance of instance enacting, but also make necessary intervention for instance running. But these operations often result in inconsistence of workflow model and workflow instances. It is quite often that inconsistency causes the workflow instance to be terminated if no corresponding adjustments could be taken. Firstly, a number of necessary constraint rules for model correctness and instance consistence are given based a workflow model—ADEPT, and then a states-based approach to monitoring and controlling is proposed, that includes a set of primitive transforming task instance from current state to next and algorithms transferring process instance from current status to future. The use of the approach guarantees both model correctness and instance consistence. Finally, completeness and consistence of the set of primitive are discussed briefly.