The literature on 443 cases of postinjection bacterial arthritis is reviewed. The utter rarity of the condition may have been overemphasized in the past. Haematogenous infection of the puncture track may be more important than the transfer of skin bacteria into the joint.