The development of effective road safety programs is dependent upon our knowledge of the various risk levels existing on the roads and highways. One of the major factors affecting road users' risk is the incidence of drivers' alcohol impairment levels. This problem is the focus of much attention in recent years, stimulating research efforts and attracting major countermeasures in Canada and elsewhere.

In response to the requirement for a better understanding of the extent and nature of the drinking-driving problem, night-time surveys of drivers' alcohol use were jointly undertaken by Transport Canada and the provinces of British Columbia, Saskatchewan and Quebec in the spring and summer of 1981. The survey operations were carried out by provincial government agencies with advice and assistance from Transport Canada in survey design, sampling estimation procedures, instrument design, data processing and analysis. An earlier paper described the survey and sampling methodologies, and provided summary results.

This paper extends earlier work and presents the results of multivariate analyses. Detailed analyses of the level of drivers' BACs (blood alcohol concentrations) cross-classified by different driver-vehicleenvironment- trip travel pattern characteristics are examined. Estimates of the distributions of drivers' BACs and their variances are computed in order to identify relationships between the distributions of drivers' BACs and trip characteristics. The knowledge gained from in-depth analyses such as these provides a basis for directing enforcement and education programs at highrisk target groups.