Aim of this study was to assess the US findings of patients with late-stage hip OA undergoing total hip arthroplasty (THA), and to associate the US findings with conventional radiography (CR) and intraoperative findings. Moreover, the inter-rater reliability of hip US, and association between the US and Oxford Hip Score (OHS) were evaluated. Sixty-eight hips were included, and intraoperative findings were available on 48 hips. Mean patient age was 67.6 years and 38% were males. OA findings—osteophytes at femoral collum and anterosuperior acetabulum, femoral head deformity and effusion—were assessed on US, CR and THA. The diagnostic performance of US and CR was compared by applying the THA findings as the gold standard. Osteoarthritic US findings were very common, but no association between the US findings and OHS was observed. The pooled inter-rater reliability (n = 65) varied from moderate to excellent (k = 0.538–0.815). When THA findings were used as the gold standard, US detected femoral collum osteophytes with 95% sensitivity, 0% specificity, 81% accuracy, and 85% positive predictive value. Concerning acetabular osteophytes, the respective values were 96%, 0%, 88% and 91%. For the femoral head deformity, they were 92%, 36%, 38% and 83%, and for the effusion 49%, 85%, 58% and 90%, respectively. US provides similar detection of osteophytes as does CR. On femoral head deformity, performance of the US is superior to CR. The inter-rater reliability of the US evaluation varies from moderate to excellent, and no association between US and OHS was observed in this patient cohort.