In this paper, we investigate non-single exponential photoluminescence decays in various disordered condensed-matter systems. For such materials, two formulas for the average lifetime of system’s excited state are commonly used in the analysis of experimental data. In many cases, the choice of formula is arbitrary and lacks a clear physical justification. For this reason, our main goal is to show that the choice of correct mathematical formula should be based on the interpretation of measured photoluminescence decay curve. It is shown that depending on the investigated system, after appropriate normalization, photoluminescence decay curve can represent either a survival probability function or a probability density function of lifetime and for this reason two different formulas for the average lifetime are required. It is also shown that, depending on luminescence quantum yield, some information on the probability density function of lifetime can be lost in the process of measurement, which results in underestimated values of average lifetime. Finally, we provide an interpretation of total decay rate distributions which are frequently obtained by phenomenological modeling of non-single exponential photoluminescence decays.