We reviewed retrospectively the outcome of the treatment by impaction grafting of periprosthetic femoral fractures around loose stems in 106 patients with Vancouver type-B2 and type-B3 fractures. Eighty-nine patients had a cemented revision with impaction grafting and a long or short stem. The remaining 17 had cemented revision without impaction grafting. Fractures treated by impaction grafting and a long stem were more than five times likely to unite than those treated by impaction grafting and a short stem (odds ratio = 5.5, 95% confidence interval (CI) 1.54 to 19.6; p = 0.009). Furthermore, those with impaction grafting and a long stem were significantly more likely to unite than those with a long stem without impaction grafting (odds ratio = 4.07, 95% CI 1.10 to 15.0; p = 0.035). There was also a trend towards a higher rate of union in those treated by impaction grafting than in those without (odds ratio = 2.69, 95% CI 0.86 to 8.45; p = 0.090). Impaction grafting is being increasingly widely used for the restoration of femoral bone stock. It can be successfully applied to periprosthetic femoral fractures but a long stem should be used to bypass the distal fracture line.