We test empirically the strategic counterpart of the Adaptive Decision Maker hypothesis (Payne, Bettman, & Johnson, 1993), which states that decision makers adapt their attention and decision rules to time pressure in predictable ways. For 29 normal form games, we test whether players adapt to tightening time constraints by reducing their information search and shifting to less computationally demanding heuristics, ultimately leading to systematic changes in choices. We specify process models of each decision rule, thereby allowing us to rank the rules by complexity on the basis of the number of elementary information processing units required to execute each rule. The frequency of decision rule use conditional on time constraints is estimated using Bayesian latent class modeling. Under time pressure, the subject pool displayed a systematic shift toward heuristics of less complexity. We observed an increase in nonstrategic decision rules (particularly Level 1) and rules that choose the social optimum at the expense of more complex rules including the Nash equilibrium. Also, subjects are more likely to not acquire information about others' payoffs, thus effectively transforming the games into nonstrategic situations. The subjects' adaptation to time pressure was efficient in the sense that it did not lead to significant payoff losses.