Random testing for drugs and alcohol has become a critical issue for a
gencies and firms who employ 'safety-sensitive' transportation workers
. The recent tightening of industry standards by the Department of Tra
nsportation in which firms operating in this industry are required to
test 25% of their employees each year at an estimated annual industry
cost of $200 million provides an incentive to evaluate the effectivene
ss of ad hoc random approaches to drug testing. In this paper we propo
se a Bayesian acceptance sampling approach for the problem of random d
rug testing in the transportation industry. The model recognizes the d
ependence of the technique on the prior distribution of users in the p
opulation and on the outcome of the test itself. The approach offers a
minimum expected total cost solution and a decision rule for testing,
based upon the optimal sampling plan derived, which may then be used
to determine future testing schedules and outcomes. The comparative co
st of sampling plans derived with the Bayesian approach are compared w
ith that obtained with a random, non-economic approach. The results sh
ow that use of an economic approach can generate savings of from 8% to
90%. The approach is applied to the Los Angeles County Metropolitan T
ransit Authority as a method of monitoring and randomly testing a popu
lation of 4000 bus drivers. In comparison with their existing approach
and utilizing cost inputs provided by the Authority, acceptance sampl
ing would allow a significant increase in the amount of testing possib
le and provide a more proactive drug testing policy toward drivers who
use drugs.