We apply the Generalized Likelihood Uncertainty Estimation (GLUE) methodology to the prediction of solute contaminant transport by studying the simulation of the development of the tracer plume in the 1983 Twin Lake tracer test. We employ a random walk particle tracking model overlaid onto a finite-element solution of the flow field in a Monte-Carlo framework. Under the GLUE methodology each simulation is assigned an initial likelihood weight, based on the goodness-of-fit to the observed tracer cloud as it passes a monitor at an early stage in its development. These likelihoods are then used to provide confidence/uncertainty limits to predictions of the future development of the 'cloud'. (from Authors)