Asset management is an increasing concern for the water and wastewater industry. Condition assessment of sewer segments is an important component of sewer asset management and relies mostly on visual inspection. Observed defects are translated into a score for each segment. Although most protocols give a segment a condition grade by comparing its score with a subjective scale of numerical values, we propose a protocol to calibrate thresholds for each asset stock. Thresholds are calculated according to two sets of parameters: overall condition of the asset stock in question (estimated by a representative sample or provided by the utility manager) and assignment-error weighting (determined by the utility manager) linked to either over-estimation or under-estimation of condition grade. This method is applied to 150 km of sewers from the Greater Lyon asset stock. Sensitivity analyses of these parameters are then implemented. Three hypotheses about overall condition of the asset stock are combined with three matrices of assignment-error weights. Both parameters influence thresholds and change the assessment of the studied segments. The synthesis of such sensitivity analyses can be used to prioritise complementary investigations.