The various technology conditions exist in different phases of product development, therefore, the products can not be treated as a population and their test data also can not be simply added together in reliability assessment. Duo to the small amount of tests conducted, the value obtained in reliability assessment is usually smaller than the actual value. This paper integrates the reliability of each phase and builds up a mathematical model for assessment of reliability growth to derive the reliability assessment equations for qualified or unqualified product. This assessment method utilizes fully the trial information from every phase to reflect the reliability growth process in product development, and provides an effective means for evaluation of reliability examination on product with multiple design phases, varied population and less test data. Meanwhile, the method is used to calculate the real reliability level of a product. The conclusion shows that the assessment result is approximately the same as the reliability level that the product reaches.