ABSTRACT. Epidemiologic indicators associated with within-farm infection of Johne's disease in dairy farms in Japan were determined through a nationwide investigation of infected farms. We assumed that subsequent detection of the disease within one year after the first detection could represent the occurrence of within-farm spread occurring before the first detection. Of 594 infected farms, 158 farms (27%) had at least one additional detection. Logistic regression analysis using epidemiologic information obtained from infected farms at the time of the first detection revealed three epidemiologic indicators associated with subsequent detection. Farms at which the first cases included cattle with clinical signs were 3.8 (95% confidence interval: 2.2, 6.8) times more likely to have additional detections than those with cattle without clinical signs. Similarly, farms where two or more cattle were detected at the time of first detection and where cattle were held in a loose housing system were 2.8 (95% CI: 1.8, 4.5) and 2.0 (95% CI: 1.1, 3.6) times more likely to have additional detections than those where only one animal was detected and a tied-up housing system was used, respectively. These epidemiologic indicators are likely important determinants in the selection of farms requiring more intensive on-farm control measures. KEY WORDS: epidemiology, Johne's disease, logistic regression model, retrospective cohort study.J. Vet. Med. Sci. 69(12): 1255-1258 Johne's disease is a chronic intestinal disease of ruminants caused by Mycobacterium avium subsp. paratuberculosis which is found throughout the world [4,12]. It affects dairy industries by reducing milk yield and slaughter values, and increasing culling rates [1,3,8,10]. In the U.S.A., for example, annual economic losses due to Johne's disease have been estimated at US$200-250 million [17].Johne's disease is a notifiable disease in Japan, and by the mid-1990s, all cases were being reported to the Ministry of Agriculture, Forestry and Fisheries (MAFF). In addition to this passive surveillance, active surveillance is now also undertaken following an amendment of the Domestic Animal Infectious Disease Control Law in 1997. All targeted cattle designated by each prefecture undergo testing for Johne's disease at least once every five years. In 2004, these two surveillance methods together detected a total of 1,118 infected cattle and 608 farms, respectively [6].Because of the nature of the Johne's disease it is difficult to detect all infected cattle by single testing only. In particular, detection of subclinically infected cattle is not easily achieved by current diagnostic methods owing to their strong dependence on the timing and proper use of diagnostic tests, as well as life stage or physiological condition of the cattle, including age or stress associated with pregnancy or calving [16,20]. As a result, infected farms in Japan are placed under regulatory monitoring with more frequent testing for a certain period determined by the respective prefecture [13,14]. All c...