OBJECTIVE Acetabular impaction grafting has been shown to be very effective, but concerns regarding its suitability for larger defects have been highlighted. We report the use of this technique in a large cohort of patients, and address possible limitations of the technique. METHODS We investigated a consecutive group of 339 cases of impaction grafting of the cup with morcellised impacted allograft bone for survivorship and mechanisms for early failure. RESULTS Kaplan Meier survival was 89.1% (95% CI 83.2 to 95.0%) at 5.8 years for revision for any reason, and 91.6% (95% CI 85.9 to 97.3%) for revision for aseptic loosening of the cup. Of the 15 cases revised for aseptic cup loosening, nine were large rim mesh reconstructions, two were fractured Kerboull-Postel plates, two were migrating cages, one was a medial wall mesh failure and one had been treated by impaction alone. CONCLUSIONS In our series, results were disappointing where a large rim mesh or significant reconstruction was required. In light of these results, our technique has changed in that we now use predominantly larger chips of purely cancellous bone, 8-10 mm3 in size, to fill the cavity and larger diameter cups to better fill the aperture of the reconstructed acetabulum. In addition we now make greater use of i) implants made of a highly porous in-growth surface to constrain allograft chips and ii) bulk allografts combined with cages and morcellised chips in cases with very large segmental and cavitary defects.