The 21-mo Roadside Testing Assessment (ROSITA) project started in January 1999 and included a literature survey of drugs and medicines that have detrimental impacts on road users' performance, an inventory of the available roadside drug-testing equipment for urine, oral fluid, and sweat, an evaluation of the operational, user, and legal requirements for roadside testing equipment in the different European Union countries, and an extensive evaluation of several devices in eight countries. On-site immunoassays were used for the detection of drugs in urine, oral fluid, and/or sweat in 2968 subjects. Police officers liked having the tools to detect drivers under the influence of drugs, and they were very creative in finding solutions to the practical and operational problems they encountered. On-site drug testing gave the police confidence, and saved time and money. Police officers had no major objections to collecting specimens of body fluids. In the majority of the participating countries, oral fluid was the preferred specimen. Some on-site urine tests (Rapid Drug Screen® [American Bio Medica], Syva® RapidTest− [Dade Behring], Dipro Drugscreen 5 [Dipro Diagnostics], Triage [Biosite Diagnostics]) yielded good results (accuracy >95%, sensitivity and specificity >90% compared with reference methods), but none had good results for all assays. The sampling of oral-fluid or sweat specimens was well accepted by drivers. The on-site tests evaluated—Drugwipe® (Securetec Detektions-Systeme AG), Cozart Rapiscan (Cozart Bioscience Ltd.), and ORALscreen™ (Avitar Inc.)—were not sufficiently reliable (accuracy between 50 and 81% in comparison with blood results). Progress is needed for sampling, duration of the test, sample volume, and reliability. From 2003 to 2005, the ROSITA-II project will evaluate the newer on-site oral-fluid devices in six European countries and five US states. [ABSTRACT FROM AUTHOR]