BACKGROUND Radionuclide arthrography (RNA) is an established technique in the evaluation of hip prostheses but there is scant literature on its role in knee prostheses and no data specifically related to unicompartmental knee prostheses. We reviewed our experience with radionuclide arthrography in total (TKRs) and unicompartmental (UKRs) knee arthroplasties. METHODS A retrospective review of 66 consecutive RNA studies that either had direct surgical correlation or arthroscopic assessment in conjunction with at least 12 months of clinical and radiological follow-up. These formed the basis of our case note review. 26 studies had a confirmed diagnosis of prosthetic loosening. RESULTS By using standardised criteria for diagnosis and an early and late imaging protocol at 30 min and 4 h, RNA had an overall sensitivity of 88% and a specificity of 88% for the diagnosis of prosthetic loosening. The individual sensitivities and specificities were 93% and 86% for TKRs and 82% and 92% for UKRs. The technique was unable to detect femoral component loosening unless a long femoral stem was present, although only 8% of episodes of prosthetic loosening did not involve the tibial component. CONCLUSIONS Radionuclide arthrography should be considered as a useful diagnostic test for the evaluation of loosening of the tibial component of knee prostheses but requires early and late imaging and close attention to detail to achieve optimal results.