Our aim was to determine the incidence of occult infection and to examine the role of ultrasound sonication of the implants in cases of presumed aseptic loosening in a prospective trial. Joint swabs, aspirates, and deep tissue samples were obtained from around the prosthesis for routine microbiology. Each prosthesis was sonicated and the sonicate examined with Gram staining and extended cultures. There were 106 joints in the study of which 54 were revised for aseptic loosening and 52 were assigned to the control revision group. There were 9 positive cultures with 8/54 positive cultures in the aseptic loosening group and 1/52 in the control revision group (p = 0.017, associated OR 47.7). We found concordant results between sonication fluid culture and conventional samples in 5/9 cultures. Preoperative inflammatory markers were not prognostic for infection. Coagulase-negative Staphylococcus was the most commonly cultured organism (7/9). Previously unrecognised infection was present in 15% of patients undergoing revision for aseptic loosening. Ultrasound sonication of the removed prosthesis was less sensitive than conventional sampling techniques. We recommend routine intraoperative sampling for patients having revision for aseptic loosening, but we do not support the routine use of ultrasound sonication for its detection.