Summary:

Outliers are a common problem occurring in a set of environmental va-riables and such the observations often disturb significantly the final classification of the quality of aquatic ecosystems. Identification of outliers can reduce various sources of error in environmental monitoring, which can be caused by a person, equipment or method as well as other random factor. This paper examines how to detect outliers by using some statistical tests. Four tests were analysed: Q-Dixon, Grubbs, Hampel and quartiled. All of them were evaluated in terms of detection sensitivity of typical outliers. The analytical dataset consisted of a monthly analysis of the various forms of nitrogen and phosphorus in the water. The study was conducted Turing the single year (2010) in two rivers in Wielkopolska region (Głomia and Mogielnica) representing different degrees of degradation.