Aims: A survey to determine the prevalence and numbers of Salmonella in beef cattle presented for slaughter at abattoirs across Australia was conducted between September 2002 and January 2003. Methods and Results: Automated immunomagnetic separation (AIMS) was used for detection and isolation of Salmonella enriched from cattle faeces. Salmonella were enumerated from positive samples using a combination of the Most Probable Number (MPN) technique and AIMS. A total of 310 faecal samples were tested, 155 were from lot-fed cattle and 155 from grass-fed cattle. Salmonella spp. were isolated from 21 (6.8%) of the cattle and the prevalence amongst grass-fed cattle (4.5%) was not significantly different to that found in lot-fed cattle (9%). Counts of Salmonella in positive faeces varied from <3 MPN g(-1) of faeces to 2.8 x 10(3) MPN g(-1) and 71% of positive samples had counts <10 MPN g(-1) faeces. There was no significant difference in the mean log(10) number of Salmonella in faeces of cattle from each production system. Conclusion: Low numbers of beef cattle were found to shed Salmonella at the time of slaughter and the prevalence and the associated faecal concentrations did not vary significantly with the pre-slaughter production system (grass or lot feeding). The faecal concentration of Salmonella in the majority of faeces was low (<10 MPN g(-1)) with few high concentrations up to 3 x 10(3) MPN g(-1), suggesting there may be a low risk of carcase contamination. Significance and Impact of the Study: Beef cattle do not appear to be a major source of entry of Salmonella into the human food chain and the quantitative information contained in this study can be used in quantitative assessments of the associated risk of human salmonellosis.