In present work we study the accelerated failure time model for survival regression data. We present various ways on how to check the model assumptions with help of both classic regression methods and martingale theory.