To ensure an optimal operation, many sewer networks are equipped with online sensors. The measurement data allows to assess the states of the sewer system not only in retrospect (referred to as offline), but also in real-time (online). The data can be processed automatically with mathematical algorithms or models to support the operation of the system directly (by real time control), or to obtain information beyond that provided by the sensors themselves. In the second case the techniques are referred to as software sensors. This work introduces software sensors and their application to derive new information from hydraulic measurements in sewer networks. They are based on conceptual models, consider uncertainties in data and model parameters, and can be used online or offline. Among the presented approaches and applications, two basic methodologies can be differentiated: The first one is referred to as reverse modelling, which denotes the estimation of system inputs from measured outputs. The second methodology is referred to as model updating, i.e. the sequential assimilation of the model to new measurements from the real system, in order to improve the estimation of model outputs in real time. Reverse modelling is used to estimate net areal precipitation on the corresponding urban catchment area from flow measurements in the sewer system. The method is applied to analyse uncertainties in areal rainfall. In combination with an error model the method is used to fill gaps in time series of measured rainfall. The second application of reverse modelling deals with the estimation of inflow to combined sewer overflow and storage structures in sewer systems. Model updating is applied to a conceptual model to simulate combined sewer overflow in real time. Data which provides only information on overflow occurrence is used to update the distribution of a model parameter.