This paper presents the statistical modelling of the monitoring data on oxygen indicators in the Rega River, collected in years of 1992-97. In the first part, the six probability distributions are examined to fit the experimental data. The results indicate that the lognormal and Gamma distributions give an adequate description for all considered data series, as measured by P-P probability plots, and a Kolmogorov-Srnirnov goodness of fit test. In the second part, the detailed statistical analysis was performed based on lognormal distribution. This study showed that the approximation of monitoring data with the appropriate probability distribution could be very useful for evaluating of water quality.