The occurrence of Mycobacterium bovis infection in cattle herds during the period 1966–92 in two geographically related areas in South-West England is compared. In one area comprising 104 km2 all badgers were systematically destroyed from 1975–81, after which recolonization was allowed; in the other, comprising 116 km2, small scale, statutory badger removal operations were undertaken from 1975 onwards where specific herds were detected with M. bovis infection. In the area with total clearance, no further incidents with M. bovis isolation occurred from 1982–92. Survival analysis and proportional hazards regression indicated that the risk of herds being identified with infection was less once badgers had been cleared from their neighbourhood, whereas it was greater in herds with 50 or more animals, and once cattle in a herd had responded positively to the tuberculin skin test, even though infection with M. bovis was not confirmed subsequently. The study provides further evidence that badgers represent an important reservoir of M. bovis infection for cattle and that badger control is effective in reducing incidents of cattle infection with M. bovis if action is thorough and recolonization is prevented.