The Danish surveillance-and-control program for Salmonella in slaughter pigs was introduced in 1995. The key element of the program is a quick and correct identification of herds with high seroprevalence. After 5 years, the classification scheme was evaluated--and a revision was made. Data from two Salmonella screenings including a total of 1902 slaughter pig herds were used. For each herd, information was available on Salmonella status based on both microbiology and serology. Based on analyses of these data, suitable changes in the scheme were identified and their effect estimated by use of data from the Danish Salmonella Database including all herds in 2000. The classification scheme has been adjusted on the following points. (1) The sampling has been simplified into 60, 75, or 100 samples per herd per year depending on herd size. This means more-precise estimates for the seroprevalence among smaller herds. (2) Herds with an annual kill <or=200 finishers will not form part of the surveillance; this leaves 1.6% of the slaughter pigs outside the surveillance scheme. (3) The cut-off for individual meat-juice samples has been reduced from OD% 40 to OD% 20--doubling the number of positive samples. (4) The results of the previous 3 months' serological samples will be weighed 0.6:0.2:0.2 (the immediate month counting three times as much as the previous months), and the weighed average is called the "serological Salmonella index" for slaughter pig herds. A herd with an increasing seroprevalence will be assigned to a higher Salmonella level more-quickly under the new scheme. (5) A herd will be assigned monthly to one of three levels. The limit between Levels 1 and 2 has been set to >or=index 40, and the limit between Levels 2 and 3 to >or=index 70. If the Danish swine producers are interested, a Level 0 may be introduced (consisting of seronegative herds as an indication of a negligible Salmonella prevalence). The classification scheme was introduced in August 2001.