In an investigation of X-ray counting circuits the fundamental quantity governing component interaction and system counting efficiency was the pulse duration associated with a single counting event. Theoretical expressions governing the behaviour of counting systems consisting of two components were derived assuming either nonself- prolonging (nsp) or self-prolonging (sp) behaviour for each component. In an experimental study, preamplifiers, amplifiers pulse height analysers and scalers were found to be sp, sp, nsp and nsp respectively. It was found that many commercial counting chains could be reduced to two-component systems consisting of an sp amplifier with pulse duration tau and an nsp pulse height analyser with pulse duration lambda . Experimental measurements of system losses for tau > lambda revealed that the efficiency was given by m/n=exp(-n tau ) where m and n are the measured and true intensities respectively. When lambda > tau , the efficiency was found to be m/n=1/(exp(n tau )+n( lambda - tau )). The range of applicability of these expressions is discussed, as are the effects of pulse amplitude shifts, baseline selection and X-ray energy. Recommendations are made concerning the measurement of deadtime, the use of the proper correction model, and equipment testing and selection.