We describe experiments to produce large quantities of the polymerase domain of E. coli DNA polymerase I for biochemical and biophysical studies. The polymerase domain derivative used in previous studies was insoluble when overproduced and tended to aggregate during purification. These problems were solved by a combination of two distinct strategies. By changing the expression system, we were able to obtain the overproduced protein in a soluble form, a necessary first step since attempts to purify the polymerase domain from the insoluble pellet were unsuccessful. The tendency of the polymerase domain to aggregate was eliminated by re-engineering the protein so as to remove both a solvent-exposed hydrophobic patch and a potentially unstructured region at the extreme N-terminus. Unlike the original construct, the re-engineered derivatives chromatographed as a single species and could be purified to homogeneity in good yield. Our experience in this study emphasizes the level of ignorance of the factors that influence protein overproduction and the need, in difficult cases, to evaluate many strategies in a semi-empirical manner.