A nation-wide Salmonella enterica surveillance and control programme was initiated in Danish finishing herds over the first quarter of 1995, In Denmark, all swine for slaughter are identifiable by a unique herd code. For each herd code, and depending on the herd's annual kill, random samples ranging from four to more than 60 swine are obtained quarterly at the abattoir. A meat sample from each pig is frozen, and meat juice (harvested after thawing) is examined for specific antibodies against S. enterica using an indirect enzyme-linked immunosorbent assay (ELISA). The ELISA combines several S. enterica O-antigens, and allows detection of antibody response after a variety of different S. enterica serovar infections. Results are transferred to a central database, which each month (based on meat-juice tests obtained in the previous 13 weeks) assigns all herds into three S. enterica infection levels: Level 1, in which the S. enterica prevalence is deemed low and acceptable; Level 2, where there is a moderate prevalence of S. enterica seroreactors (from > 50% in the smallest to > 10% in the largest herds); Level 3, in which S. enterica seroreactor prevalence is clearly unsatisfactory (> 50% for most herd sizes). Irrespective of Salmonella level, all herds receive a monthly update on the current results of the S. enterica test results. IF a herd is categorized in Level 2 or 3, it must receive an advisory visit by a practising veterinarian and a local swine extension specialist, and certain management hygiene precautions must be taken. If a herd is categorized in Level 3, the finishers from the herd must additionally be slaughtered under special hygiene precautions. This is supervised by the veterinary authorities. During 1995, 604 000 samples were tested for S. enterica, corresponding to 3.0% of the total kill. In December 1995, 15 522 herds (representing > 90% of the national production) were categorized into one of the three levels: 14 551 herds (93.7%) in Level 1; 610 herds (3.9%) in Level 2; 361 herds (2.3%) in Level 3. The proportion of serologically positive meat-juice samples collected during 1995 ranged from a mean of 2.9% in smaller herds (101-200 swine slaughtered per year) to 6.1% in relatively large herds (more than 5000 swine slaughtered per year).