Derived from the Karush–Kuhn–Tucker (KKT) optimality conditions of utility maximization, the multiple discrete–continuous extreme value (MDCEV) models are extensively used in the field of time use. This is because they are grounded in a framework that maximizes underlying utility, showcasing appealing behavioral attributes and computational simplicity. However, the existing MDCEV framework and its sophisticated versions all employ a utility specification for time consumption that increases monotonically. This approach does not take into account the possibility of non-monotonic preferences in decisions related to time use. In this paper, we propose, for the first time, a new closed-form MDCEV model with flexible marginal utility and utility functions, which breaks the assumption of the strictly increasing utility specification. Retaining some basic features of the traditional MDCEV model, the new formulation also incorporates the behavioral foundation of non-monotonic preferences. The improvement is achieved by introducing an additional linear utility component into the traditional utility expression. We also develop a simple and computationally-efficient forecasting algorithm for the new flexible model. Monte Carlo simulation experiments are then conducted to validate the model and the forecasting algorithm. In empirical analysis, a -profile of the proposed formulation is applied to analyze weekend time use pattern of non-commuters in Shanghai, China, exhibiting its capability to accommodate non-monotonic preferences based on behaviorally-plausible interpretations. What’s more, a comparative study between the new flexible model and the traditional model reveals differences in their behavioral foundations, with our new flexible model displaying superior performance in data fitting and activity duration forecasting.