On the basis of a modified Constant scoring system, we compared outcomes for 16 patients who underwent latissimus dorsi transfer as a salvage reconstruction for a failed prior rotator cuff repair with outcomes for 6 patients who underwent a primary reconstruction for an irreparable cuff defect. There was a statistically significant difference in Constant score between groups, which measured 55% for the salvage group compared with 70% for the primary group (P <.05). Poor tendon quality, stage 4 muscle fatty degeneration, and detachment of the deltoid insertion each had a statistically significant effect on the Constant score (P <.05). Late rupture of the tendon transfer occurred in 44% of patients in the salvage group compared with 17% in the primary group at a mean of 19 months postoperatively. Rupture had a statistically significant effect on the Constant score, which declined by a mean of 14% (P <.05). We conclude that salvage reconstruction of failed prior rotator cuff repairs yields more limited gains in satisfaction and function than primary latissimus dorsi transfer.