(devices that are considered accurate enough for the results to be used as evidence in law courts) are a means of substantially increasing breath-testing activity. Though used in most high-income countries, they are not currently widespread elsewhere. This greatly limits the ability of many countries to respond effectively to the problem of drink- driving. The deterrent effect of breath-testing devices is to a large extent dependent on the legislation governing their use (126). Police powers vary between countries, and include the following: — stopping obviously impaired drivers; — stopping drivers at roadblocks or sobriety checkpoints and testing only those suspected of alcohol impairment; — stopping drivers at random and testing all who are stopped. The following components have been identifi ed as being central to successful police enforcement operations to deter drinking drivers (128): • A high proportion of people tested (at least one in ten drivers every year but, if possible, one in three drivers, as is the case in Finland). This can only be achieved through wide-scale application of random breath testing and evidential breath testing. • Enforcement that is unpredictable in terms of time and place, deployed in such a manner so as to ensure wide coverage of the whole road network and to make it diffi cult for drivers to avoid the checkpoints. • Highly visible police operations. For drinking drivers who are caught, remedial treatment can be offered as an alternative to traditional penalties, to reduce the likelihood of repeated offending. Random breath testing and sobriety checkpoints Random breath testing is carried out in several countries, including Australia, Colombia, France, the Nordic countries, the Netherlands, New Zealand and South Africa. The use of sustained and intensive random breath testing is a highly effective means of reducing injuries resulting from alcohol impairment. In Australia, for instance, since 1993 it has led to estimated reductions in alcohol- related deaths in New South Wales of 36% (with one in three drivers tested), in Tasmania of 42% (three in four tested) and in Victoria of 40% (one in two tested) (126). An international review of the effectiveness of random breath testing and sobriety checkpoints found that both reduced alcohol-related crashes by about 20% (149). The reductions appeared to be similar, irrespective of whether the checkpoints were used for short-term intensive campaigns or continuously over a period of several years. A Swiss study has shown that random breath testing is one of the most cost-effective safety measures that can be employed, with a cost– benefi t ratio estimated at 1:19 (150). In New South Wales, Australia, the estimated cost–benefi t ratio of random breath testing ranged from 1:1 to 1:56 (126, 151, 152). Similarly, economic analyses on the sobriety checkpoint programmes in the United States estimated benefi ts totalling between 6 and 23 times their original cost (153, 154). Mass media campaigns It is generally accepted that enforcement of alcohol impairment laws is more effective when accompanied by publicity aimed at: — making people more alert to the risk of detection, arrest and its consequences; — making drinking and driving less publicly