Utilities are faced with the challenge of how to manage their assets cost-effectively while providing safe and reliable services to their customers (Marlow et al., 2007). A small number of utilities have completely inspected their sewer networks. Therefore, the use of an asset stock's sample to calibrate deterioration models and to study scenarios about the future seems mandatory. This sample should reflect the asset stock's characteristics in the best manner. However, authors have calibrated deterioration models without paying attention to the impact of used samples on their outcomes. The main scopes of this article are then to: 1) draw a representative sample of an asset stock, 2) provide a reliable estimation of a specific property of the asset stock from this sample and 3) study the impact of calibration sample on the outcomes of a multivariate model. We show that the calibration of deterioration models depends heavily on the characteristics of the used sample.