Polymorphism of drugs has been the subject of intense interest in the pharmaceutical industry for over forty years. Although identical in chemical composition, polymorphs differ in bioavailability, solubility, dissolution rate, chemical and physical stability, melting point, colour, filterability, density, flow properties, and many other properties. The difference in solubility is particularly important for pharmaceuticals, as it can affect drug efficacy, bioavailability and safety. Despite significant investment in processes to find all the possible polymorphs of active pharmaceutical ingredients (APIs), new polymorphs can suddenly appear without warning. Polymorphs tend to convert spontaneously from less stable to more stable forms, and, therefore, it is best to discover and characterize the stable form as early as possible. Ideally the most stable polymorph will be found while the drug candidate is still in the discovery process, so that this is the form used for subsequent testing. The most stable polymorph will be the least soluble and solubility may be a limiting factor in the efficacy of the API. Despite the huge importance of polymorphism in the properties of materials, however, there is no method that can produce all the stable polymorphs of a compound, or even one that can provide confidence that the most stable polymorph has been obtained. Here we describe a new method, `potentiometric cycling for polymorph creation (PC)2', which is able to generate the most stable polymorph in aqueous solution. This new method has been applied to sulindac, a non-steroidal anti-inflammatory drug, which also shows promise in anticancer treatment, producing two polymorphs of this API, including a new more stable one. By adjusting the conditions, this method is able to produce either polymorph exclusively.