A failure analysis of interference screw fixation was performed to test the hypothesis that bovine and/or elderly human cadavers are appropriate models for bone-patellar tendon-bone anterior cruciate ligament (ACL) reconstruction fixation studies. Failure mode is an important criterion for validating experimental models. The bovine, young human, and elderly human failure loads were 799 +/- 261 N, 655 +/- 186 N, and 382 +/- 118 N, respectively, and the failure modes were 75%, 69%, and 30% tissue failures, respectively. The similarities between the bovine and young human models in failure loads and failure modes indicate that bovine models are appropriate for ACL reconstruction fixation studies. The statistically significant differences between the young human and elderly human models in failure loads and failure modes indicate that elderly human cadavers are not an appropriate model for ACL reconstruction fixation studies. The differences in failure modes are consistent with previous studies using elderly human cadavers in which the predominant failure mode was bone block pullout. The tissue failures observed in the bovine and young human models contradict previous studies suggesting fixation strength is the weakest link in bone-patellar tendon-bone ACL reconstruction. Results of linear regression modeling showed statistically significant correlations between insertion torque and failure load (R2 = 0.44, P < 0.0001) and interference (defined as the screw outer thread diameter minus the tunnel-bone block gap) and insertion torque (R2 = 0.18, P = 0.003) when data from all models was combined. Results for the bovine model multiple regression showed a statistically significant regression of insertion torque (linear) and interference (quadritic) versus failure load (R2 = 0.56, P = 0.02). Regression slopes for screw diameter (P = 0.52) and gap size (P = 1.00) were not statistically significant. These results indicate that insertion torque and interference are independent predictors of failure load and should be included in future interference screw studies in addition to bone block dimensions, tunnel size, gap size, and screw diameter. Clinicians may consider using insertion torque and interference as indicators of postoperative graft fixation regarding rehabilitation decisions.