The findings show that LVA did not exhibit the expected sensitivity to the presence of deception, truth, and/ or high or low stress in the speech samples that composed the experimental database for this project. The observed true positive and false positive rates varied by the particular evaluation team and by the particular analysis conducted. The raw data and all statistical analyses suggest only chance-level performance by the LVA. There are some common alternative interpretations of the testing results that could be used to argue that the device was not adequately tested in a laboratory study; for example, the negative result could reflect limitations in the protocols used in the development of the speech database. It could be argued that the stress shifts documented for the speech samples were not of a comparable magnitude to those induced in situations outside of the laboratory. In such cases, the “real-world” levels of stress might be higher and more measurable than the psychological stress produced in a laboratory setting on a college campus. A counter-argument is that if the database, which was collected under highly controlled conditions, contained inadequate levels of “real-world” stress, then very low false positive rates (near zero) would have been observed. In fact, high false positive rates were the norm across all sets of speech materials and across both teams of operators. Future research should focus on differences in results due to the skill and experience of the interrogator in interaction with the LVA. Subjects were 24 males and 24 females drawn from a diverse population. 1 table, 2 figures, and 60 references