There are a wide variety of defect (bug) detection techniques available to the practising software engineer, but which one should they use? We are told for example, that code reviews and inspections outperform the various testing techniques, but tight schedules and budgets often constrain the developer to applying just one technique. So what is the engineer to do - inspect the software and not test it, or vice-versa? This talk describes work done to determine which are the most effective defect detection techniques and examines the impact of combining different techniques, and the consequences of not doing so.