A mathematical model for predicting the growth response in patients with Turner syndrome who received growth hormone (GH) therapy was developed by analysing data from KIGS, the Pharmacia & Upjohn International Growth Database. A model for year 1 of GH therapy explained 46% of the variability of the growth response, with GH dose being the most important of the predictors of height velocity. In years 2-4 of therapy, height velocity during the previous year was the most important predictor, suggesting that an individual's initial response to GH may determine the height outcome of treatment. Additional predictors of height velocity in years 1-4 of GH therapy included age (negative), weight SDS and additional treatment with oxandrolone. The predictions in all 4 years were highly accurate, as indicated by the low error SDs. However, relatively low predictive power (R) during years 2-4 of treatment suggests the models are missing other parameters that would explain more of the variability of the growth response. These growth prediction models could help clinicians to design individualized treatment regimens, provide realistic expectations of therapy outcomes, and adjust treatment on the basis of detected differences between observed and predicted height velocities.