This paper operationalizes and tests approaches to identify market segments for rail freight services and measures the importance that customers attach to rail service attributes (i.e., transport cost, time, frequency, reliability, and safety). The approach is based on choice-based conjoint analysis in which heterogeneity is captured by means of latent-class analysis. The research is novel in several respects. First, it addresses the diverse valuation of service preferences by shippers who use rail transport. Second, besides estimates for rail users who contract for transport services, the analysis also arrives at new estimates for forwarders as immediate clients for rail services. Third, in addition to the conventional random utility maximization (RUM) model, the paper discusses trials with a random regret minimization (RRM) model and a hybrid RUM-RRM model. Finally, the research produces unique values for China, one of the largest rail transport markets in the world.