When considering the credibility assessment of suspects in police interviews, there is a wealth of information in practitioner and empirical literature, suggesting a range of verbal behaviours that are supposedly inked to deception and truthfulness. However, much of the practitioner literature lacks empirical validation, whilst empirical research often suffers from low ecological validity. The present thesis sought to address these shortcomings by analysing the verbal and paralinguistic behaviours of 133 suspects in real life police interviews in relation to high volume offences. An initial study showed that suspects (n= 26) classified as truth-tellers, subtle liars and outright liars exhibited a number of significant verbal and paralinguistic differences, but also found that such a conceptualisation of suspects limited use for practitioners. A second study examined the transcribed initial accounts of 85 suspects, and tested the validity of a number of credibility criteria suggested by the police literature. Suspects were classified as guilty/admission, guilty/omission, guilty/denial and innocent/denial. The results showed that although there were a number of trends in the data, only a few statistically significant differences differentiated the groups. The third study (n = 107) examined suspects' verbal behaviour throughout the interview, with suspects classified as guilty/admission, guilty/partial admission, guilty/denial and innocent/denial. In addition, suspects' responses were categorised as having either low or high incriminating potential. The results indicated a number of behavioural differences that were particularly pronounced during responses with high incriminating potential. A final study found that certain suspect and case characteristics (e.g. strength of evidence, previous convictions) had predictive validity in relation to guilty suspects' (n = 85) likelihood to admit or deny. Overall, the findings suggest that a number of verbal and paralinguistic behaviours previously linked to deception are more likely to be reflective of a 'guilty' conscience', and that there is a need for more research in the applied setting to understand the complex interactions within the interview room.