The Danish Salmonella Surveillance-and-Control Programme in finisher pigs includes both herd and carcass surveillance. Herd surveillance consists of serological testing of meat-juice samples and classification of herds into three Salmonella seroprevalence levels. At the abattoirs, carcass swabs from five pigs are collected daily and analysed as a pooled sample to evaluate the Salmonella carcass prevalence. This study aimed to investigate factors associated with Salmonella carcass prevalence in Denmark. A total of 20,196 pooled carcass swabs collected in 23 Danish abattoirs were included in the analysis. A multilevel logistic regression model was used taking into account the two-level data structure (abattoir, carcass pool) and adjusting the parameter estimates to the random variation at the abattoir level. Study results indicated that carcass contamination was mainly influenced by the probability that at least one pig contributing to the pool was seropositive, the log-transformed number of seropositive pigs delivered to the abattoir on the same day and weekday. No other factors were found to be significant (P>0.05). Large reductions in the number of seropositive pigs delivered to slaughter are unlikely to result in large reductions of the Salmonella carcass prevalence, unless the number of seropositive pigs can be kept below approximately 200. On average, individual Salmonella carcass prevalence can be kept below 1% by keeping a Salmonella input to the abattoir below approximately 50 seropositive pigs. Variation between abattoirs suggested that improved hygiene practices in some of the abattoirs would reduce the Salmonella carcass prevalence further.