Bootstrapping technique is being used as an alternative method for internal validation of a developed model. It involves a large number of samples with replacement from the original sample. Other methods for internal validation like data splitting, and cross validation are also used but bootstrap, i.e., the re-sampling method gives less biased and more consistent results than the others [5, 7, 12]. The main aim of this study is to use Cox Proportional Hazard (Cox PH) regression analysis to develop child survival model and to validate the model for its predictive accuracy. Around 2.1 million child deaths occur every year in India with mortality rates varying from state to state. To utilize important factors of child survival and achieve the millennium development goals on under-five mortality reduction, separate child survival models are required for each state. Before utilizing the factor affecting child survival from these developed models to health management and policy makers, predictive accuracy of a model is extremely essential which can be judged through its validation. Total 4873 children of first to fifth birth order born during the 4 years preceding the National Family Health Survey (NFHS), 1992-93, Uttar Pradesh (UP), India was considered [14]. Breastfeeding, father�s education, religion/caste, immunization, antenatal care and premature were found important factors for child survival. Shrinkage coefficient was 92%, indicating an excellent model calibration. Somer�s D rank correlation was �0.58, indicating good internal validity. Results are found good enough to describe the predictive accuracy for the model.