Initially, models of chemical systems displaying oscillatory behavior were judged successful if they could show how such behavior was even possible. Recently, however, reaction mechanisms for chemical oscillators have been subjected to more stringent experimental tests. I examine strategies for model testing that flow from theoretical considerations, in particular, the types of feedback relations between chemical species required to produce oscillatory behavior in mechanistic models. These theoretical considerations allow chemists to work around important practical considerations such as an inability to measure certain species in the experimental system. These sorts of testing strategies illuminate chemists’ desiderata for a “good model.”