The purpose of this study was to compare the diagnostic accuracy for the detection of infection between the culture of fluid obtained by sonication (SFC) and the culture of peri-implant tissues (PITC) in patients with early and delayed implant failure, and those with unsuspected and suspected septic failure. It was hypothesised that SFC increases the diagnostic accuracy for infection in delayed, but not early, implant failure, and in unsuspected septic failure. The diagnostic accuracy for infection of all consecutive implants (hardware or prostheses) that were removed for failure was compared between SFC and PITC. This prospective study included 317 patients with a mean age of 62.7 years (9 to 97). The sensitivity for detection of infection using SFC was higher than using PITC in an overall comparison (89.9% versus 67%, respectively; p < 0.001), in unsuspected septic failure (100% versus 48.5%, respectively; p < 0.001), and in delayed implant failure (88% versus 58%, respectively; p < 0.001). PITC sensitivity dropped significantly in unsuspected compared with suspected septic failure (p = 0.007), and in delayed compared with early failure (p = 0.013). There were no differences in specificity. Sonication is mainly recommended when there is implant failure with no clear signs of infection and in patients with delayed implant failure. In early failure, SFC is not superior to PITC for the diagnosis of infection and, therefore, is not recommended as a routine diagnostic test in these patients. Cite this article: Bone Joint J 2013;95-B:244–9