We validated a prototype cuff-free device for noninvasive estimation of blood pressure (BP). The system assumed a linear relation between BP values and the inverse of arterial blood pulse transit time, measured as time interval between the R wave on the electrocardiograph and the onset of the peripheral pulse wave on a finger plethysmogram. Thirty-three healthy subjects were analyzed at rest and during increasing stress exercise. To estimate subject-specific linear model parameters, the system was calibrated ad personam with reference to BP measures obtained by a cuff sphygmomanometer. High correlation values (R(2) = 0.89 and 0.78 for systolic and diastolic BP, respectively) and differences consistent with clinical requirements (mean discrepancy of -0.058 and -0.25 mm Hg; 95% confidence interval of -13.0 to +12.9 mm Hg and -11.3 to +10.8 mm Hg, for systolic and diastolic BP, respectively) were observed between device and reference measurements. Calibration parameter stability and accuracy level were confirmed in a midterm evaluation, 30 days after calibration. These findings suggest the suitability of the device for noninvasive BP monitoring and its potentiality for clinical applications. Improvements can be achieved by further investigation of the calibration procedure and sensor placement.