Annual nutrient loads have been estimated for Lake Balaton over three decades. Tributaries may transport about half of the loads into the lake. The contribution of diffuse sources may reach two thirds of the total load. Biweekly/monthly water quality monitoring on small inflows (0.01 m3/s-0.3 m3/s range) results in a high uncertainty of load estimates. This paper evaluates the degree of uncertainties by using analytical expressions of sampling theory. Load-flow relationships were derived for five streams and annual total phosphorus load was predicted by four load estimation methods. A seasonal regression model, based upon the evaluation of historical set of observed phosphorus loads, appeared best to refine load estimates on small inflows. Correction frequently led to load estimates that exceeded uncorrected loads by a factor of two to three. Since the dynamics of the watercourses determined the errors of load estimates, stratified sampling is needed to decrease the uncertainties.