The torque output of the Nobelpharma electric torque controller was determined using a special setup on an Instron test machine. The devices were held in a vice and oriented so that activation of the drivers caused a pure torsion effect. Measured torque levels generally varied significantly between individual devices. Further, significant differences generally existed between the nominal torque levels and the corresponding measured values. Torque application errors varied from target values by 1% to 165%. For the clinician to deliver appropriate torque levels, these data indicate that a simple means for chairside calibration and adjustment should be available. Parole chiave: biomechanics, implants, prosthodontics, torque control