This paper will discuss the problem of improving reliability prediction on the basis of analysis advantages and disadvantages of current situation in theoretical and practical aspects of reliability prediction for industry. This relates to different areas of engineering. Predicting is inaccurate when it is based on information obtained from using traditional approaches of accelerated life testing (ALT) data where the degradation (failure) process differs substantially from the product’s degradation process during service life under real world conditions. The paper will consider currently used reliability prediction methods through the analysis of publications and author’s thinking. It will be analyze why considered reliability prediction approaches not successful in industry that leads to many recalls, less reliability, durability, maintainability, and profit, and higher life cycle cost of the product that might be. The author will introduce a new approach to reliability prediction, which eliminated considered negative aspects of current situation with reliability prediction.