Patients with infected arthroplasties are normally treated with a two-stage exchange procedure using polymethylmethacrylate bone cement spacers impregnated with antibiotics. However, spacers may act as a foreign body to which micro-organisms may adhere and grow. In this study it was hypothesised that subclinical infection may be diagnosed with sonication of the surface biofilm of the spacer. The aims were to assess the presence of subclinical infection through sonication of the spacer at the time of a second-stage procedure, and to determine the relationship between subclinical infection and the clinical outcome. Of 55 patients studied, 11 (20%) were diagnosed with subclinical infection. At a mean follow-up of 12 months (interquartile range 6 to 18), clinical failure was found in 18 (32.7%) patients. Of the patients previously diagnosed with subclinical infection, 63% (7 of 11) had failed compared with 25% (11 of 44) of those without subclinical infection (odds ratio 5.25, 95% confidence interval 1.29 to 21.4, p = 0.021). Sonication of the biofilm of the surface of the spacer is useful in order to exclude subclinical infection and therefore contributes to improving the outcome after two-stage procedures.