People have shown sensitivity to variance in studies in which variance has been provided separately from other statistical information, but not in other studies in which variance must be derived from raw data. However, such studies typically test people’s sensitivity to variance via probability judgments: participants are asked to make judgments based on how confident they are that sample means are representative of a population. In this study, we instead investigate whether people are able to use variability when making likelihood judgments: participants determined from which of two possible populations a sample was more likely to have been drawn. Choices were influenced by variance, even when controlling for sample size, base rate, and the absolute difference between sample means and population µs.