A method and apparatus to evaluate the performance of an implanted medication delivery device such as pumps and ports is described. In the case of a pump, the system measures ambient pressure and obtains two pressure measurements from the implanted device. One of these measurements is device reservoir pressure and the other is based on a disturbance pulse injected into the system at a point downstream of the reservoir. Based on those measurements and input calibration data, an expected flow rate is determined. The expected flow rate together with device refill data is in turn used to determine a normalized flow rate for the in-vivo device. Determinations are then made as to the performance of the reservoir, its pump and the condition of the outlet catheter without explant of the device. In the case of a port, the system measures the catheter resistance based on decay time of a disturbance pulse.