By default, PROC MCMC discards all observations that have missing values before carrying out the posterior sampling. This corresponds to the option MISSING=CC, where CC stands for complete cases. PROC MCMC does not automatically augment missing data. However, you can choose to model the missing values by using MISSING=AC. Given this option, PROC MCMC does not discard any missing values. It is up to you to specify how the missing values are handled in the program. You can choose to model the missing values as parameters (a fully Bayesian approach) or assign specific values to them (multiple imputation). In general, however, the handling of missing values largely depends on the assumptions you have about the missing mechanism, which is beyond the scope of this chapter.