We investigated factors associated with persistence of different Salmonella serovars in buildings housing laying hens in Great Britain using survival analysis. A total of 264 incidents of Salmonella detection occurring between July 1998 and August 2007 in 152 houses were recorded. For incidents involving Salmonella Enteritidis (SE), both the rodent score of the house and the type of house were positively associated with persistence. For non-SE serovars, only the type of house was associated with persistence. Persistence of SE in the houses was longest (>15 months) in step-cage and cage-scraper houses when high levels of rodents were present, and lowest in non-cage and cage-belt houses. We estimated that 42% (95% CI 23.3-63.1) of SE incidents may be cleared during the lay period, and this was related to elimination of rodents from the houses. From January 2009, EU legislation will ban the sale of fresh eggs from SE-positive and S. Typhimurium-positive flocks over their remaining lifespan. If infection is eliminated from such flocks, they would cease to represent a public health risk.