NIST provides calibration service for instruments that customers in turn use to calibrate other instruments. Often during their useable lifetime, these instruments are calibrated and recalibrated several times. Ideally, if an instrument has not degraded too badly, then using the prior calibration data should improve the calibration estimate. A dynamic calibration model is set up that takes into account not only the comparison data taken at the calibration session, but also all prior calibration data. Dynamic calibration is a recursive Bayesian algorithm that is adapted from the dynamic linear model. Using data from the calibration of a pressure transducer, it is shown that using the prior data does improve recalibration accuracy.