Of 283 patients who had a massive allograft of bone, an infection developed in thirty-three (11.7 per cent). To assess the frequency and identify the co-morbid and predisposing factors of this devastating complication, we compared demographic data for the infected and non-infected patients. Comparison of mean age, type of graft, anatomical site of the procedure, and stage of the tumor yielded no significant differences. Multiple-regression analysis of a subgroup of eighty-two patients who had a distal femoral graft showed a correlation between infection and factors that are associated with more extensive surgery (more loss of bone, soft tissue, or skin) or with multiple operations. Approximately 30 per cent of the patients who had an infected allograft had no co-morbid or predisposing factors that could be statistically correlated with an increased risk for infection. Gram-positive organisms were the most common cause of infection, with twelve infections (36 per cent) being due to Staphylococcus epidermidis. Six patients had a single gram-negative organism and nine had mixed flora. The final result in the thirty-three patients who had an infected allograft was poor compared with that of the over-all series and of the uninfected patients. Twenty-seven infected allografts (82 per cent) were considered to be failures of treatment because amputation of the limb or resection of the graft was required to control the infection.(ABSTRACT TRUNCATED AT 250 WORDS)