Real uncertainties in the parameter values of conceptual models used in hydrology sometimes remain after calibration. While this is unacceptable, uncertainty analysis can provide useful information on the structural properties of the model. This information may then be used to modify the model or establish the limits of usefulness of the measured data. This paper focuses on the estimation and interpretation of the parameter uncertainties in such cases. The standard estimation method, based on the second-order approximation of the shape of the objective function in the vicinity of the optimum parameter set, provides an inaccurate description of these uncertainties. Instead a global approach, consisting of scanning the hypercube of the possible parameter values, is shown to give a good overview of these uncertainties. This gives not only a good general view of the shape of the objective function, but a good basis for interpretation as well. The efficiency of the method is illustrated using a four parameter urban stormwater quality model. The main conclusion is that the lack of measured data and its relative redundancy cause a significant interaction between the parameters, hindering calibration. Another example is given of something being reported by a growing number of authors: that data available in hydrology are only sufficient to support the development of models with limited complexity.