The purpose of this study was to estimate personality and mental health through behavior data measured by acceleration and voice intensity sensors. Traditionally measuring methodologies require huge amount of time and resources for its operation. Techniques that attempt to classify and measure psychological states require acknowledgment of its dynamic behavior, and the issues intrinsic to the use of self report inventories. This research conducted experiments in real-life settings, minimizing intrusiveness to participants. A methodology for estimating personality and mental health in the work place was proposed. Sensor-based behavior analysis provided an unobtrusive and time-efficient mechanism to estimate psychological states through measurement of a selected set of behaviors. This methodology’s objective was to demonstrate existing correlations between estimated behavior and assessed personality and mental health. The results showed significant correlations between behavior and all personality and mental health states studied except for openness. This research provides insights into the analysis of personality and mental health states in the working settings. More broadly, the methodology proposed in this research provides implications for the development of recognition systems that will facilitate the attainment of personal and collective goals, which proves highly useful in today’s increasingly technicized societies.