We describe a method to decipher the complex inter-relationships between metabolite production trends and gene expression events, and show how information gleaned from such studies can be applied to yield improved production strains. Genomic fragment microarrays were constructed for the Aspergillus terreus genome, and transcriptional profiles were generated from strains engineered to produce varying amounts of the medically significant natural product lovastatin. Metabolite detection methods were employed to quantify the polyketide-derived secondary metabolites lovastatin and (+)-geodin in broths from fermentations of the same strains. Association analysis of the resulting transcriptional and metabolic data sets provides mechanistic insight into the genetic and physiological control of lovastatin and (+)-geodin biosynthesis, and identifies novel components involved in the production of (+)-geodin, as well as other secondary metabolites. Furthermore, this analysis identifies specific tools, including promoters for reporter-based selection systems, that we employed to improve lovastatin production by A. terreus.