Purpose It has previously been suggested that failure to achieve 70% metaphyseal fit with the Austin Moore prosthesis predisposes to subsidence and failure of the prosthesis secondary to stem loosening. We performed a radiological review to test this hypothesis, and determine whether a correlation existed between head size and metaphyseal fit of the uncemented Austin Moore hip prosthesis. Methods Post-operative radiographs of 102 uncemented Austin Moore hemiarthroplasty patients were retrospectively reviewed. Three observers blinded to the head size of the prosthesis, measured the percentage metaphyseal fit of each stem, using a previously published technique. Results Eighty-six females and 16 males were included in this study. Eighteen of the radiographs were excluded for incorrect implantation, narrow stem insertion, and un-interpretable post-operative radiographs. Implants with a head size of 42 mm had an average metaphyseal fit of 84.5%. Those with head size 54 mm had only a 56% metaphyseal fit. The result was strongly associated with an R value of −0.93. Conclusion We have shown an inverse correlation exists between prosthesis head size and metaphyseal fit. Austin Moore prostheses with head sizes greater than 49 mm are likely to be loose from the moment of insertion. This poses a particular risk in men. In our study, only one male required prosthesis with a head less than 49 mm. In these patients, use of an alternative prosthesis such as a bipolar uncemented or cemented hemiarthroplasty is recommended.