We reviewed the plain radiographs, bone scans and hip aspiration results of 54 patients with painful hip arthroplasties which had been explored surgically, to compare the results of the investigations with the operative findings. For acetabular loosening, the sensitivity and specificity of bone scanning were 87% and 95%, with an accuracy of 90%: for serial plain radiography sensitivity was 95%, specificity 100% and accuracy 97%. For femoral component loosening, bone scan sensitivity was 85%, specificity 100% and accuracy 89%: the sensitivity of plain radiography was 100%, with specificity 92% and accuracy 98%. Technetium bone scanning did not provide additional information with regard to loosening and is not necessary in the routine investigation of a painful hip arthroplasty. Serial pain radiography is the most effective method of detecting loosening, and bone scanning is useful only when radiography is inconclusive with regard to loosening or infection.