Turbidity monitoring is a valuable tool for operating sewer systems, but it is often considered as a somewhat tricky parameter for assessing water quality, because measured values depend on the model of sensor, and even on the operator. This paper details the main components of the uncertainty in turbidity measurements with a special focus on reproducibility, and provides guidelines for improving the reproducibility of measurements in wastewater relying on proper calibration procedures. Calibration appears to be the main source of uncertainties, and proper procedures must account for uncertainties in standard solutions as well as non linearity of the calibration curve. With such procedures, uncertainty and reproducibility of field measurement can be kept lower than 5% or 25 FAU. On the other hand, reproducibility has no meaning if different measuring principles (attenuation vs. nephelometry) or very different wavelengths are used.