This paper deals with a comparison of the pollution measurement accuracy between real time pollution sensors (especially Suspended Solids - S.S.- determination by turbidimetry) and standard samples analysis on raw wastewaters. The two methods are used in commonly favourable conditions. Considering separate measurement errors, the S.S. random errors of the two methods are in the same order of magnitude (about ± 25% for a 90% confidence interval), but sampler/analyses lead to a systematic underevaluation of about 17%. These rather high figures for the sampling/analyses method are due to the cumulation of samples compounding and preservation errors. The determination of the samples collection error when estimating pollution loads is based on a set of 12 rainfall events registered during a 5 month study on a combined sewer overflow. Due to an optimized sampling strategy, the use of a 24 bottle sampler shows an average S.S. load underevaluation of only 5%, compared with the continuous estimation of the pollution sensor considered as a reference. Real time pollution sensors can, therefore, be considered as convenient systems for field experiments where particulate pollution is of special interest, which is the case for runoff waters. Besides which, they are particularly suitable for the control and automation of sanitary transport and treatment equipment, being easily integrated into computerized systems.