This study proposes a new method to quantify mental workload (MWL) automatically, without interfering with the operator's primary task performance. An unobtrusive Tobii eye tracker recorded eye movements while participants were engaged in a cognitively demanding N-back task. Original algorithms automatically analyzed the eye data, detected specific eye deviation movements called saccadic intrusions (SIs), and automatically quantified the eye deviation accounted for SIs. This SI measure was strongly correlated with the task difficulty levels in the N-back tasks and with pupil diameter. This indicates that the SI measure appeared to reflect MWL and may be used as a measure of MWL.