This article presents an extensive study to select the theoretical probability distribution that represents the concentrations of major constituents from 35 wastewater treatment plants located in Brazil, as well as to verify the influence of the adopted treatment technology on the data distributions. Due to the lack of a solid basis for selecting the distributions that best fit the data, various distributions had to be tried and the choice was based on statistical tests and visual techniques, comprising a total of 3,444 tests undertaken. While comparing suitability of five distributions (normal, lognormal, gamma, exponential and rectangular) for analysis of data (influent and effluent biochemical oxygen demand, chemical oxygen demand, suspended solids, nitrogen, phosphorus, thermotolerant coliforms and flow), it was observed that the lognormal distribution was the most suitable, which is in agreement with the findings from other research studies. This conclusion requires a different position from the one currently adopted when analysing plant performance, in which symmetry of the data is generally implied, which has been shown not to be the case with the large array of data sets investigated.
Keywords: lognormal behaviour, probability distribution, wastewater constituents