The DRINK code is a 2D, biogeochemical transport code developed as a research tool to simulate the long term evolution of near surface LLW disposal sites and to generate gaseous and liquid source terms for far field studies. The code was recently upgraded to provide a more generic modelling tool with wider application to radionuclide migration scenarios. During the development of this code, the Generalised Repository Model (GRM), an integrated strategy has been employed to ensure the production of a fully tested, verified and quality assured product. This strategy is based around a code development protocol with three main components: quality assurance and documentation, verification and realism testing. Realism testing includes both peer review and model testing, with the latter including: experimental test cases; natural and anthropogenic analogues; field observations and finally uncertainty and sensitivity analysis. This paper describes the successful application of the protocol to the development and testing of the GRM code with specific emphasis upon verification and realism testing.