Methods of calibrating circuits for measuring partial discharges and radio-interference (r.i.) voltages are discussed. It is shown that discharge-measuring circuits, whether using a peak-measuring discharge detector or an r.i. meter, are best calibrated by applying a charge-quantity pulse calibrator of known repetition frequency in parallel with the test sample. The calibration of circuits for measuring r.i. voltages is more complex, and certain anomalies arise when a sine-wave signal generator is used to calibrate the circuit in terms of voltage input. It is established that, for a pulse-repetition rate of 100 pulse/s, it is possible to relate the results obtained from a conventional peak-reading discharge detector to those from a quasipeak r.i. meter; e.g. for a meter having 60? input impedance and a bandwidth of 9 kHz, the relationship is: 1 ?V is equivalent to 2.6pC. At repetition rates other than 100 pulse/s, it is shown that theoretically the maximum error involved by using the above relationship is 6 dB between 25?2000 pulse/s. Practical measurements confirm that the error does not exceed ±6dB over a wide range of pulse sizes and repetition frequencies. Comparative measurements have shown that the conventional r.i. meter is approximately an order of magnitude less sensitive than a tuned-circuit discharge detector. Methods of improving the sensitivity of the former are described. It is concluded that for many applications either a discharge detector or an r.i. meter will enable both discharge magnitude and r.i. voltage to be determined with sufficient accuracy, considering the erratic behaviour of pulses in practice.