The synchronization of autonomic arousal levels and other physio-logical responses between people is a potentially important component of work team performance, client-therapist relationships, and other types of human interaction. This study addressed several problems: What statistical models are viable for identifying synchronization for loosely coupled human systems? How is the level of synchronization related to psychosocial variables such as empathy, subjective ratings of workload, and actual performance? Participants were 70 undergraduates who worked in pairs on a vigilance dual task in which they watched a virtual reality security camera, rang a bell when they saw the target intruder, and completed a jig-saw puzzle. Event rates either increased or decreased during the 90 min work period. The average R2 values for each person were .66, .66, .62, and .53 for the linear autoregressive model, linear autoregressive model with a synchronization component, the nonlinear autoregressive model, and the nonlinear autoregressive model with a synchronization component, respectively. All models were more accurate at a lag of 20 sec compared to 50 sec or customized lag lengths. Although the linear models were more accurate overall, the nonlinear synchronization parameters were more often related to psychological variables and performance. In particular, greater synchronization was observed with the nonlinear model when the target event rate increased, compared to when it decreased, which was expected from the general theory of synchronization. Nonlinear models were also more effective for uncovering inhibitory or dampening relationships between the co-workers as well as mutually excitatory relationships. Future research should explore the comparative model results for tasks that induce higher levels of synchronization and involve different types of internal group coordination.