This article presents a new theory of subjective probability, called support theory, which challenges the classical assumption of extensionality. The theory posits that different descriptions of the same event can lead to different probability judgments. Experimental evidence supports the theory's predictions: judged probability increases when the focal hypothesis is unpacked and decreases when the alternative hypothesis is unpacked. Judged probabilities are complementary in binary cases and subadditive in general cases, contradicting classical and revisionist models. Subadditivity is more pronounced in probability judgments than in frequency judgments and is enhanced by compatible evidence. Support theory proposes that probability judgments are based on the explicitness of event descriptions rather than on the events themselves. It introduces a ratio scale of support, where the judged probability of an event is determined by the relative support of the focal and alternative hypotheses. The theory is nonextensional, meaning that events with the same extension can have different probabilities depending on their descriptions. The theory is extended to ordinal judgments and to the assessment of upper and lower probabilities. It is applied to both laypeople and experts in evaluating uncertain events. The theory is supported by experimental evidence showing that subadditivity is a fundamental aspect of human judgment. It also accounts for the influence of memory and attention on probability judgments, as well as the role of salience in affecting support. Support theory is contrasted with Bayesian and revisionist models of belief. While Bayesian models assume additive probability measures, support theory uses nonadditive set functions. The theory is also compared with Shafer's theory of belief functions, which assumes extensionality and superadditivity. Support theory, however, assumes subadditivity for implicit disjunctions and additivity for explicit ones. The theory is validated through various experiments, including studies on causes of death, suggestibility, and expert judgments. These studies show that subadditivity is not limited to novices but also holds for experts. The theory is further supported by studies on probability judgments for uncertain quantities and on binary complementarity. The findings suggest that subadditivity is a fundamental principle of human judgment, and that probability judgments are influenced by the explicitness of descriptions and the salience of outcomes.This article presents a new theory of subjective probability, called support theory, which challenges the classical assumption of extensionality. The theory posits that different descriptions of the same event can lead to different probability judgments. Experimental evidence supports the theory's predictions: judged probability increases when the focal hypothesis is unpacked and decreases when the alternative hypothesis is unpacked. Judged probabilities are complementary in binary cases and subadditive in general cases, contradicting classical and revisionist models. Subadditivity is more pronounced in probability judgments than in frequency judgments and is enhanced by compatible evidence. Support theory proposes that probability judgments are based on the explicitness of event descriptions rather than on the events themselves. It introduces a ratio scale of support, where the judged probability of an event is determined by the relative support of the focal and alternative hypotheses. The theory is nonextensional, meaning that events with the same extension can have different probabilities depending on their descriptions. The theory is extended to ordinal judgments and to the assessment of upper and lower probabilities. It is applied to both laypeople and experts in evaluating uncertain events. The theory is supported by experimental evidence showing that subadditivity is a fundamental aspect of human judgment. It also accounts for the influence of memory and attention on probability judgments, as well as the role of salience in affecting support. Support theory is contrasted with Bayesian and revisionist models of belief. While Bayesian models assume additive probability measures, support theory uses nonadditive set functions. The theory is also compared with Shafer's theory of belief functions, which assumes extensionality and superadditivity. Support theory, however, assumes subadditivity for implicit disjunctions and additivity for explicit ones. The theory is validated through various experiments, including studies on causes of death, suggestibility, and expert judgments. These studies show that subadditivity is not limited to novices but also holds for experts. The theory is further supported by studies on probability judgments for uncertain quantities and on binary complementarity. The findings suggest that subadditivity is a fundamental principle of human judgment, and that probability judgments are influenced by the explicitness of descriptions and the salience of outcomes.