Transformers are one of the most important and high-cost components in energy transmission with an approximately 30 years expected lifetime. Because of this reason, their absence according to a fault may cause blackouts and extreme financial losses. Besides, faults which may occur in the active part which includes the core and the windings of a transformer can only be repaired in the manufacturer's facilities. Therefore, the repair time is extremely long and also the winding renewal is not cost effective. Lifetime estimations of transformers which are popular solutions for the mentioned problem can be achieved with the help of developed algorithms by online tracking of the variations in specific transformer parameters. The accuracy of loss of life estimation is greatly affected by the correct calculation of hot-spot temperature. In this study, hot-spot temperature calculation is improved by using experimental measurement data obtained with fiber optic temperature sensors for the loss of life time estimation of a natural cooling cylindrical winding distribution transformers. The present work focuses specifically on reducing the error rate without increasing input data. With the help of proposed method, more accurate lifetime calculations are done by reducing the error from 50 up to 0.8%.