An improved theory of the instrumental corrections applied to absolute radiometers, with the exception of the effect of sensitivity variations across the receiver surface, is derived. After a brief review of the historical emergence of correction factors and finally of a formal correction theory, as formulated by Geist [ 131, some shortcomings of t h s theory are pointed out. Based on a set of strict definitions and the conservation of energy principle a revised theory is derived and a possible experimental procedure for the determination of the various correction factors is described. The new theory also takes into account the influence of deviations from linear response, which to a certain extent are present in all practical absolute radiometers. The ratio of the mean sensitivities for radiant and electrical heating is expressed in terms of either the equivalent or non-equivalent lieat losses. By analyzing the one-dimensional heat flow through a representative model of an absolute radiometer the equivalent and non-equivalent losses are identified. At the same time a formal proof for the conception of the equivalence of the heat loss through the heater wires is provided. All correction terms in the new theory are treated as functions of the applied power and can also be determined as such. In conclusion both the scope and the limitations of the theory are discussed.