The application of uncertainty analysis requires appropriate statistics that capture the degree of uncertainty in model forecasts. The prediction of activity-based models of travel demand relates to (a) aggregate performance indicators, (b) origin–destination tables and corresponding traffic flows, (c) individual space–time trajectories, and (d) the sequence of activities that are conducted during the day. Because measures to quantify uncertainty in the sequence of activities have not been developed yet, the aim of this paper is to propose a method to measure model uncertainty in predicted activity travel sequences. The proposed method involves generating predicted activity travel patterns for a set of model runs and quantifying the uncertainty in the sequential information embedded in these patterns by calculating the average effort across all possible pairs of predicted sequences to align these multidimensional sequences. The effort is quantified in a multidimensional extension of the Levenshtein distance. Because computational costs may become prohibitive in large-scale applications of this method, several heuristic approaches are suggested and examined. Results indicate that the suggested heuristics (a) can represent uncertainty in predicted activity travel sequences quite well, (b) tend to approximate the calculated uncertainty on the basis of all possible sequences, and (c) do not however necessarily produce asymptotically more accurate results. Implications of these findings are discussed.