Although the principle of bounded rationality seems more realistic for formulating formal models of individual choice behaviour than traditional decision-outcome-based discrete choice models, existing studies have some limitations: (1) applications focus on the noncompensatory nature of the models and largely ignore the factor-selection process; (2) heterogeneity of heuristics in reaching a decision is insufficiently studied; (3) the choice of decision strategy is rarely modeled formally. A modeling approach that simultaneously deals with these issues is suggested. Factor thresholds are used as the mechanism for factor selection and representation, resulting in a set of activated and nonactivated factor states. Under the assumption of stochastic contextual effects, the model automatically generates heterogeneous decision heuristics, including the conjunctive, disjunctive, and lexicographic rule. Mental effort and risk perception are assumed to influence the evaluation and choice of heuristics. The concept of preference tolerance is used to predict the probability of selecting a particular heuristic under different decision contexts. The modeling approach is illustrated using the go-home decision of pedestrians in a shopping street in China as an example.