The successful engineering of secondary metabolite production relies on the availability of detailed computational models of metabolism. In this brief review we discuss the types of models used for synthetic biology and their application for the engineering of metabolism. We then highlight some of the major modeling challenges, in particular the need to make informative model predictions based on incomplete and uncertain information. This issue is particularly pressing in the synthetic biology of secondary metabolism, due to the genetic diversity of microbial secondary metabolite producers, the difficulty of enzyme-kinetic characterization of the complex biosynthetic machinery, and the need for engineered pathways to function efficiently in heterologous hosts. We argue that an explicit quantitative consideration of the resulting uncertainty of metabolic models can lead to more informative predictions to guide the design of improved production hosts for bioactive secondary metabolites.