The actual performance of an Eppley pyrgeometer is compared to the desired theoretical performance. Several systematic errors are identified and evaluated in detail. The three most significant errors identified are due to 1) battery voltage uncertainties, 2) nonlinearity of circuitry at extreme temperature and 3) differential heating of the instrument. The elimination of the error due to differential heating is found to be essential to the successful calibration of the instrument. A pyrgeometer laboratory calibration technique is described. Pyrgeometer measurements made from aircraft are shown to have potential errors as large as 60 W m−2. These errors, however, do not significantly affect the net radiation provided the upward and downward facing pyrgeometers are at the same equilibrium temperature, and may be largely eliminated by making accurate temperature measurements of the KRS-5 dome and the cold junctions of the thermopile. The corrections considered in this paper not only reduce the absolute errors but significantly decrease the transient response of the instrument. The feasibility of using an empirical expression to correct errors due to solar heating is also demonstrated for aircraft measurements.