Human cases [94] of clinically failed, cemented, total hip prostheses requiring surgery were reviewed to determine the accuracy of preoperative plain radiography, culture of aspirated fluid, arthrography and bone scanning. When radiopaque cement was used to embed the prosthesis, plain radiography was highly accurate in detecting a loose femoral component, less so in detecting a loose acetabular component. Culture of aspirated fluid was accurate in diagnosing infection. A positive arthrogram identified loosening with good accuracy; a negative arthrogram did not reliably exclude loosening. 99mTc bone scans frequently differentiated loosening from loosening with infection. The suggested sequence of diagnostic tests is plain radiography followed by bone scanning. If the bone scan shows diffuse augmented uptake, culture of aspirated fluid followed by arthrography is indicated.