This paper presents an experimental study to investigate *strategic sophistication*, the extent to which behavior in games reflects attempts to predict others' decisions and consider their incentives. The study uses a series of 18 normal-form games with various patterns of iterated dominance and unique pure-strategy equilibria, allowing subjects to search for hidden payoff information while recording their searches. The experimental design aims to minimize learning and repeated-game effects to better understand how subjects' decisions are determined and to predict deviations from equilibrium decisions. The theoretical and econometric framework is based on a mixture model where each subject's behavior is determined by one of nine types, including strategic and nonstrategic types. The types are defined to capture general principles of decision-making and are selected for their appropriateness and separation of implications for decisions and information search. The analysis focuses on the relationship between information search and decisions, using a procedural model of decision-making. The results show that subjects' decisions are heterogeneous, with a high frequency of strategic types but most subjects' sophistication is better described by boundedly rational strategic types. Information searches are even more heterogeneous, and the econometric analysis confirms the behavior suggested by the analysis of decisions alone, with significant differences. The most frequent types are Naïve and $L2$, each accounting for nearly half the population, and a third type, $D1$, accounting for 10-20%. The analysis suggests a simple view of subjects' behavior, with two types comprising 67-89% of the population and $D1$ accounting for 10-20%. The experimental design includes a Baseline treatment, an "Open Boxes" treatment, and a "Trained Subjects" treatment to control for learning and cognitive limitations. The games are chosen to separate strategic and nonstrategic types and to stress-test equilibrium predictions. The analysis of decisions and information search provides insights into how subjects' cognitive processes influence their strategic behavior.This paper presents an experimental study to investigate *strategic sophistication*, the extent to which behavior in games reflects attempts to predict others' decisions and consider their incentives. The study uses a series of 18 normal-form games with various patterns of iterated dominance and unique pure-strategy equilibria, allowing subjects to search for hidden payoff information while recording their searches. The experimental design aims to minimize learning and repeated-game effects to better understand how subjects' decisions are determined and to predict deviations from equilibrium decisions. The theoretical and econometric framework is based on a mixture model where each subject's behavior is determined by one of nine types, including strategic and nonstrategic types. The types are defined to capture general principles of decision-making and are selected for their appropriateness and separation of implications for decisions and information search. The analysis focuses on the relationship between information search and decisions, using a procedural model of decision-making. The results show that subjects' decisions are heterogeneous, with a high frequency of strategic types but most subjects' sophistication is better described by boundedly rational strategic types. Information searches are even more heterogeneous, and the econometric analysis confirms the behavior suggested by the analysis of decisions alone, with significant differences. The most frequent types are Naïve and $L2$, each accounting for nearly half the population, and a third type, $D1$, accounting for 10-20%. The analysis suggests a simple view of subjects' behavior, with two types comprising 67-89% of the population and $D1$ accounting for 10-20%. The experimental design includes a Baseline treatment, an "Open Boxes" treatment, and a "Trained Subjects" treatment to control for learning and cognitive limitations. The games are chosen to separate strategic and nonstrategic types and to stress-test equilibrium predictions. The analysis of decisions and information search provides insights into how subjects' cognitive processes influence their strategic behavior.