Summary Annual intravenous administration of zoledronic acid is used in the treatment of osteoporosis. A mathematical model was developed to predict bone mineral density up to 2 years after two annual doses of zoledronic acid from the early values of a bone resorption marker in osteoporosis patients. Introduction The measurement of bone mineral density (BMD) has been used as a surrogate marker instead of the observation of incident fractures to detect the efficacy of treatment. However, this method requires a long time to obtain significant changes. On the other hand, bone resorption markers respond to bone resorption inhibitors within a few weeks. Therefore, the aim of this study was to develop a mathematical model predicting long-term BMD after two annual doses of zoledronic acid (ZOL) using the early response of a bone resorption marker in osteoporosis patients. Methods The model was constructed using 3410 tartrate-resistant acid phosphatase 5b (TRACP-5b) serum concentrations and 1146 lumbar spine (L2-L4) BMD values from 306 patients with primary osteoporosis. A mathematical model was developed to describe the time-dependent profiles of TRACP-5b and BMD. Results The percentage changes from baseline of the BMD (%BMD) at up to 2 years were predicted from patients’ baseline BMD and baseline and 12-week TRACP-5b values by the model obtained. The simulated 90% prediction interval almost covered the observed %BMD distribution at each time point, and the predictions were comparable to the observed %BMD. Conclusions This is the first model to predict BMD for up to 2 years following two annual doses of ZOL using patients’ background characteristics and the early response of TRACP-5b. This model allows us to inform patients at the initial stage of ZOL treatment of their predicted response to treatment.