A subset of a real data set from a national survey on alcohol use and driving in Canada (original sample of 6,457) is used to illustrate an ad hoc analysis with missing data on multiple response variables. A complete-case analysis initiates this strategy, determining variables that may be deleted without losing effects of interest. By such a deletion, the number of complete observation vectors may very well increase. Also illustrated are two straightforward imputed means analyses. All illustrations are given in the context of predictive discriminant analysis. As discussed, the ad hoc strategy may be applicable to other multivariate contexts. Four tables illustrate steps in the analysis. (Contains 11 references.) (SLD)