<p>Mean concentrations of pollutants and nutrients are frequently used in the classification of water bodies under the European Commission Water Framework Directive. Turbidity can be used as a proxy for several particle-associated substances. We used three years of sensor turbidity data in each of 10 Nordic streams to assess the uncertainty in the annual arithmetic mean concentration of turbidity when sampling weekly, fortnightly, or monthly. On average, weekly sampling produced an uncertainty of 17±9% for all streams, whereas monthly sampling increased this to 40±19%. We then developed a multiple linear model that shows how the uncertainty in annual mean turbidity increases with decreasing sampling frequency, decreasing catchment size, and increasing proportion of agricultural land in the catchment. As an example, a river of 800 km² catchment area and 20% agricultural land will need 29 samples a year to give a mean annual turbidity within 10% uncertainty, but a stream of 10 km² catchment area and 80% agricultural land will need 95 samples to achieve the same uncertainty. Turbidity was a suitable proxy for total phosphorus in seven of the 10 streams. The range of the annual mean total phosphorus concentration was compared with the environmental objective (good water quality) in each river. The results highlight that when the estimated mean total phosphorus concentration is close to the concentration of the environmental objective, an infrequent sampling strategy can lead to false water quality classification. .</p>