Past empirical studies comparing the performance of trained interpreters and untrained bilinguals in police interviews have focused on the accuracy of the propositional content. Little is known about the relative skill of these two groups. These skills were empirically tested in a simulated police interview. A total of 100 English-Spanish bilinguals and trained interpreters in the greater Sydney area interpreted while professionally-trained actors role-played the detective and suspect parts. Convergent results from quantitative analyses and discourse analysis of 17 illustrative excerpts confirmed that interpreters with specialized training outperformed the untrained bilinguals. Ad hoc interpreters were less confident, used inappropriate colloquial and powerless speech styles, failed to explain their role or establish ground rules that all statements would be interpreted, used first and second person, breached ethical guidelines on impartiality, and did not interpret all utterances. The trained interpreters were rated as more trustworthy, confident, likeable and knowledgeable than their untrained counterparts. Although skills in managing turn taking and ethical dilemmas fall outside the scope of propositional accuracy, they are integral to professional interpreting and distinguished the performance of the two participant groups. Implications of the findings for interpreter training and police policy and practice in investigative interviews are discussed.