 Overdispersion is a common problem in ecological and evolutionary biological data sets, often resulting from missing covariates, aggregation of data points, or excess frequencies of zeros. Ignoring this issue can lead to biased parameter estimates and false conclusions about hypotheses of interest. Observation-level random effects, OLREs, have been proposed as a solution to this problem, but their efficacy has yet to be rigorously tested. In this study, we used simulations to investigate how well OLREs perform in different scenarios of overdispersion. We found that OLREs were effective in reducing bias in parameter estimates when overdispersion was caused by random extra Poisson noise or aggregation of data points. However, they failed to reduce bias in zero-inflated data, and in some cases increased it. Furthermore, there was a positive relationship between the magnitude of overdispersion and the degree of bias in parameter estimates. These results suggest that OLREs provide a simple and robust method for dealing with overdispersion in count data, but should be applied judiciously depending on the type of overdispersion present. This article was authored by Xavier A. Harrison. We are article.tv, links in the description below.