Rear-end crash is the predominant crash type in work zones. This study employs a random parameters logit approach with heterogeneity in means and variances to investigate factors affecting the injury severity of rear-end crashes in work zones using the data collected in North Carolina. The injury severity is divided into three categories (i.e., injury, possible injury, and property damage only). The relationships between the severity outcomes and drivers’, environmental, roadway, work zone, and crash characteristics are explored. Likelihood ratio tests are conducted to examine the temporal stability of estimation results between different time periods. A significant temporal instability is found between 2010 and 11 and 2012–13. A comparison of the marginal effects using different approaches is performed as well. Compared with the random parameters logit model only, the random parameters logit approach with heterogeneity in means and variances produces the best fitting results.