The Sentinel blood culture system was used for the analysis of 657 specimens from infected prosthetic joints and blood cultures (83 from prosthetic joints and 574 from standard blood cultures). The positivity rate was similar for specimens from prosthetic joints and blood cultures (18% compared with 14%). However, there was an unacceptable rate of false positive results with specimens from prosthetic joints (58% compared with 8%). This high false positivity rate was due to (i) prolonged incubation and (ii) the lack of blood in these specimens. It is therefore recommended that the Sentinel system should only be used for the initial seven days of incubation of specimens taken from prosthetic joints. Further incubation should take place in a standard incubator and a terminal subculture performed after 21 days.