The differential diagnosis between asepting loosening or prosthetic joint infection is not always easy. Tc-99m-labelled leucocyte scans, frozen section and histology can help recognise doubtful cases. We report the experience of the Unit for Infectious Diseases and Septic Orthopaedics of the ASL-2 Liguria, Italy, with a Tc-99m-labelled leucocyte scan and intraoperative frozen section to choose the best therapeutic approach: one-stage or two-stage exchange or arthrodesis-arthroplastica. All cases underwent histology and intraoperative cultures to confirm the diagnosis, and the effectiveness of the approach was evaluated at follow up after 18 months. From January 2011 to December 2012, 36 patients were evaluated (21 hip and 15 knee arthroprosthesis). The Tc-99m-labelled leukocyte scan was positive in 31 and negative in 5 patients. Frozen section was negative in 7 patients. Five of them were patients with a negative Tc-99m-labelled leucocyte scan and were treated successfully with one-stage exchange, even if, in one of them, Enterococcus faecalis was isolated at replacement and suppressive antibiotic treatment was needed. The other 31 patients were treated with arthrodesis arthroplasty (3 patients) or a two-stage exchange. In this group the Tc-99m-labelled leucocyte scan was positive in all patients and the frozen section was positive in 29/31 cases with 6% false negative. Histology was positive in 27/31 with 13% of false negative. The sensitivity and specificity value was respectively 90% and 100% in the frozen section, 84% and 100% in histology. Cultures were positive in 23/31 cases. Patients subjected to two-stage exchange were evaluated again during prosthesis replacement but the results of the Tc-99m-labelled leucocyte scan and histology showed unclear results more frequently: the Tc-99m-labelled leucocyte scan was positive in two cases, the frozen section in three and histology in seven in spite of positive culture in three cases and one relapse in a patient with a frozen section and histology positive but negative culture. In the second step sensitivity and specificity were respectively 67% and 96% in the frozen section, 75% and 88% in histology. Finally, our experience suggests the utility of nuclear and histological tests during diagnostic work-up for differential diagnosis of aseptic loosening or prosthetic joint infection. The same tests produce more uncertain data when performed in two-stage exchange during the second step.