The application of an optical fiber sensor on the identification of hand postures by force myography technique is reported. The transducers are comprised of fiber microbending elements attached to the user forearm, allowing the assessment of muscular activities related to the hand movements. Next, the output light intensities are detected, and then correlated to the performed postures by means of artificial neural networks. The methodology was applied on the identification of 5 postures, yielding a 96.7% accuracy considering the average of 4 subjects. Finally, the sensor was utilized for controlling a virtual prosthetic hand implemented on V-REP environment, based on the event driven finite-state approach, providing the real-time execution of predefined grasp patterns.