Aims. Infection following total hip or knee arthroplasty is a serious complication. We noted an increase in post-operative infection in cases carried out in temporary operating theatres. We therefore compared those cases performed in standard and temporary operating theatres and examined the deep periprosthetic infection rates. Patients and methods. A total of 1223 primary hip and knee arthroplasties were performed between August 2012 and June 2013. A total of 539 (44%) were performed in temporary theatres. The two groups were matched for age, gender, body mass index and American Society of Anesthesiologists grade. Results. The deep infection rate for standard operating theatres was 0 of 684 (0%); for temporary theatres it was eight of 539 (1.5%) (p = 0.001). Conclusion. Use of a temporary operating theatre for primary hip and knee arthroplasty was associated with an unacceptable increase in deep infection. We do not advocate the use of these theatres for primary joint arthroplasty. Cite this article: Bone Joint J 2017;99-B:917–20