In the 1980s rule-based systems became very popular in the domain of expert systems. It soon became apparent, that large rule-based systems causes enormous maintenance problems, because of the lack of separation between domain knowledge an control strategy. Rules are declarative, weakly structured, difficult to manage and maintain and should be applied only in local contexts and with limited use. For large rule bases the user can not be sure if the problem is completely covered by the rules, modifications often result in unwanted consequences. For application in business process management (BPM) rules became popular again, but the known insufficiencies still remain. Today it seems to be quite ignored that techniques using rules have strucural drawbacks that limit their application significantly. We present a constraint-based approach to enhance process models with additional knowledge. Constraints allow compact modeling of decision processes to inference specific values in process models. Furthermore constraints may be used as mechanism for quality assurance (QA). Constraints also have a declarative paradigma but avoid the lack of separation between domain knowledge and control strategy. Constraint solvers used as black box by a process engine will merely compute an output based on the given domain knowledge and return these as new input for the process engine. The constraint solver will give control as soon as possible back to the process engine. The process engine exclusively has to decide how to proceed. So constraints support a process engine without competing for the control strategy.