Introduction: As a form of regression analysis, spline regression can be an appropriate method to show the relationship between predictor and response variables in nonlinear models. Spline regression is not usually used in medical sciences. This study aims to compare pattern of changes in HbA1c level among diabetic patients using two regression models together with a clustering technique. Methods: This study was done in 2013 and randomly included 834 patients who referred to Hamzeh clinic. 95 patients were involved to diabetic mellitus and entered to study. Data was collected based on demographic questionnaire and laboratory results. Data were clustered and then analyzed using simple linear regression and spline regression with and without clustering. Results: Although no significant statistical relationship was detected between HbA1C factor changes and age of diabetic patients by using Pearson correlation coefficient and linear regression, Regression spline showed statistical relationship between two factors. Conclusion: The results indicated that amount of HbA1C is indirect position with age of patients with HbA1C ≥ 14. Using a clustering technique as well as spline regression analysis to fit curves to data, changes in dependent variables can be explained better by independent variables, and a better estimate can be made compared with other linear models.