Abstract
Food fraud is of high concern to the food industry. A multitude of analytical technologies exist to detect fraud. However, this testing is often expensive. Available databases detailing fraud occurrences were systematically examined to determine how frequently analytical testing triggered fraud detection. A conceptual framework was developed for deciding when to implement analytical testing programmes for fraud and a framework to consider the economic costs of fraud and the benefits of its early detection. Factors associated with statistical sampling for fraud detection were considered. Choice of sampling location on the overall food-chain may influence the likelihood of fraud detection.