Dealing with uncertainties is a crucial aspect of people’s everyday lives. We study how people form and update subjective probability estimates of uncertain events. People often do not judge the probability of uncertain events according to the rules of logic and probability theory. Instead, their judgments can often be better described by cognitive judgment strategies. Although these strategies often violate the rules of probability theory, they nevertheless might often lead to good decisions. We compare several models in their ability to capture the processes behind people’s behavior when having to assess and update probabilistic information. For instance, when people have to access the probability of how often it snows during winter, they could rely on a sample of snowy days in the past winter. Of course this sample will contain a sampling error, so that the subjective probability will only to some extent predict the objective probability. In comparing different cognitive models, we aim at identifying the process that describes people’s probability judgments in such error-prone environments. We will test cognitive models such as similarity models, exemplar models, and the weighted average model against Bayesian models. In probability updating, people form an initial belief about the probability of an event and then receive additional information that can be used to update the initial belief sequentially. In such situations, the dilution effect is often observed in that people revise their initial probability judgment by additional non-diagnostic information that they should better ignore. We will explore which model captures people’s behavior best. In particular, we will explore whether the cognitive process that leads to the conjunction effect also leads to the dilution effect. In general, this project aims for a better understanding of how people use different pieces of information when they have to assess the probability of uncertain events.