In most countries, it is not compulsory to be trained to work as an interpreter in community settings. A comparison across jurisdictions reveals that different requirements exist, from a simple self-evaluation of language competence, to passing a certification or accreditation test. Even in countries where certification or accreditation systems exist, such as the USA and Australia, there is no legislation to prevent any bilingual from working as interpreter. In legal settings, this situation has led to a lack of clear guidelines regarding interpreter recruitment, and to many examples of incompetent interpreting that have impacted legal outcomes. Little research has been conducted to systematically assess the value of training on interpreter performance. This paper presents results of a live experimental study conducted in Sydney, New South Wales, Australia, showing significant differences between the performance of trained interpreters and untrained bilinguals in simulated police interviews. The study is one of a few to compare performance based on interpreter background, using a large sample and a sophisticated method to assess performance.