The resolution and performance of scintillation counters used as energy spectrometers are limited by several factors due to the various physical processes involved. The spread in output pulse magnitude when detecting monoenergetic particles is, in the ideal case, due to the statistical fluctuations in the number of photoelectrons produced at the photomultiplier cathode and collected by the first multiplying stage. In practical usage contributions to the pulse spread may be due in particular to events in the scintillator and in the secondary emission processes in the multiplier. A critical discussion of the above factors is given and also of the problem of non-linear variation of phosphor response with energy and nature of the incident radiation.