In clinical studies, different types of outcomes (multivariate responses) can be observed from the same subject as we studied in Chapter 7. In this chapter, we consider data analysis for multivariate responses where at least one response is time-to-event. For example, we may have bivariate responses, where one outcome is repeated-measure response and the other outcome is time-to-event. Then, these outcomes are correlated because they are observed from the same subject. Joint modeling has been widely studied (Henderson et al., 2000; Ha et al., 2003; Rizopoulos, 2012) because a separated analysis, ignoring the inherent association between the outcomes from the subject, can lead to a biased result (Guo and Carlin, 2004). An unobserved random effect can be used to account for the association among multivariate outcomes. For the analysis of such dataset, the h-likelihood approach is very effective.