Externally bonded reinforcement with fiber-reinforced polymer (FRP) laminates is a popular method for repairing and strengthening reinforced concrete structures. Failure due to premature debonding poses the challenge of accurately estimating the FRP-to-concrete bond capacity. Such a challenge, premature debonding, exhibits the significance of accurately estimating the FRP-to-concrete bond capacity. Therefore, many lap shear tests have been conducted to evaluate the behavior and capacity of FRP-to-concrete bonds. Based on these tests and theoretical studies, numerous bond capacity estimation models have been proposed. These models, however, do not have the same levels of accuracy and performance. In this study, the accuracy and performance of available models for bond capacity are evaluated based on various criteria. For this purpose, 31 bond capacity prediction models were chosen. A database containing 345 test specimens that were used in none of the models was collected. The models were initially scored through the four methods of log-likelihood, Euclidean distance ranking, Bayesian factor, and variance reduction. The final ranking was then determined by applying the results of these methods to a multicriteria decision-making process using the data envelopment analysis technique. The results show that most models ranging from 1st to 10th are the models employed by strengthening provisions or based on fracture mechanics. The models’ coefficient calibration tests have a considerable effect on the ranking results. The effective bond length has a significant impact on the accuracy and score of the models and is the most important parameter.
