Random roadside checks show that the percentage of people driving under the influence of alcohol appears to be declining, but many weekend drivers test positive for drug use.

The findings come from the latest roadside survey by the National Highway Traffic Safety Administration based on breath, saliva, blood samples and questionnaires taken from randomly selected drivers in 300 locations around the United States. In 1973, 7.5 percent of drivers had a blood alcohol concentration of .08 or higher. (A level of .08 is above the legal limit in all 50 states.) In the latest survey, the percentage of people driving above the legal alcohol limit had fallen to 2.2 percent.

For the first time, the roadside survey also used screening methods to detect marijuana, cocaine and prescription drugs. The survey found that 16.3 percent of nighttime weekend drivers tested positive for drugs. Nearly 9 percent had used marijuana, whereas nearly 4 percent tested positive for cocaine and a similar number had used prescription drugs. The drug tests only indicate the presence of the drug in the body and don’t indicate when the drugs were used or whether the driver was impaired.

The survey data showed that men were more likely to be impaired by alcohol than women. Drivers were most likely to be legally drunk between 1 a.m. and 3 a.m., compared with daytime or evening hours. The vehicles most likely to be operated by drunk drivers were motorcycles and pick-up trucks.

The survey data were collected in 2007 from roadside locations throughout the country. Drivers were selected at random and waved off the road to a survey location by police officers, but the drivers were approached by interviewers who were not police officers. The drivers were assured that the survey was voluntary and anonymous. Of the 11,000 randomly selected drivers, about 90 percent agreed to give breath samples and 70 percent agreed to give saliva samples, said Jeff Michael, associate administrator for research and program development at the National Highway Traffic Safety Administration.

The interviewers used extra incentives to encourage participation in the survey. Drivers were given $10 for saliva samples and $50 for blood samples. When a driver refused to take part in any of the testing, they were then offered $100 as an added inducement.

Only a few hundred drivers refused to take part in the survey. While that may suggest those drivers were impaired by alcohol or drugs, the numbers were small enough that they were unlikely to have a large effect on the data. In addition, because the same methods were used during each survey, the results are a useful indicator of driving impairment trends from year to year.

Because the survey was anonymous, readings from breath, saliva and blood samples weren’t immediately available to interviewers. However, if a driver appeared to be impaired, the interviewers attempted to obtain a readable breath sample. Drivers who appeared impaired or who were confirmed to be impaired weren’t arrested, but they also weren’t allowed back on the road. Instead, they were allowed to call for a ride, driven home by fellow passengers, offered a ride by the researchers themselves or even offered a hotel room.

“They went to great lengths to prevent these people from driving home,” Mr. Michael said. “It was not an enforcement stop. The important thing was we didn’t want to allow anybody back into traffic that appeared to be impaired.”

Copyright 2009 The New York Times Company