Every year, British Telecom's (BT's) level of investment in software is growing. Not only is the level of investment high, but there is a great deal of reliance on this software to retain BT's competitive edge against stiff competition. The software is, therefore, likely to be working in a commercially critical environment where the cost of failure is high both in terms of lost revenue and good will. Techniques are available, at various levels of maturity, which can help to provide added assurance against software failure. The first part of the paper relates specifically to the ESPRIT COSMOS project and work within BT on validating measures gathered using the COSMOS tool at the source code level. The second part of the paper describes how the results of this project fit within the wider context of software assessment, and outlines some of the techniques that are currently being used within BT or are being developed. The paper concludes with a brief description of what may be viable in the next few years.>