There are many methods and standards to predict reliability. Stress based standards are widely used. Also, most of the companies perform accelerated life tests and analyze their test data with statistical distributions. However, every standard and statistical distribution gives different reliability predictions and these predictions differ from the field values. The reason for this is that, stress based standards and accelerated life tests may not consider every type of stress factors and failure mechanisms seen in the field. The main stress factors mentioned in stress based standards are tem perature, voltage and power dissipation. In addition, the main stress factors used in accelerated life tests are temperature, relative humidity, voltage and vibration. However, those stress factors are not the only failure reasons of the products in the field. For example, ESD, Inrush Current, Voltage Dips-Interruptions-Variations, Lightning, Loose Plugs etc. can cause failures in the field. Therefore, there should be a parameter which can express these failure mechanisms and this new parameter should be combined with traditional prediction methods to predict the failure rate and reliability of the product more accurately. In this paper, determining reliability by failure rate estimation with a new parameter in R&D phase is introduced. This new parameter is obtained by applying a set of electrical, environmental and mechanical tests in R&D phase, before mass production. This set of tests is created as it should be able to simulate different stress factors and failure mechanisms faced in the field. These tests can also be approval and validation tests at both board and product level. A scoring point is given to every test according to the severity of the test. The severity of the test can be decided by analyzing similar projects' field returns. A test which gives more information about failures will have higher scoring points. At the end, the total point of the tests is obtained. On the other hand, a losing point is given to every failure which is found during testing according to its severity. The severity of the failure can also be decided by analyzing similar projects' field returns. After all tests are performed, total losing points will be calculated. The new parameter mentioned above is de fined as the ratio of losing points to total test points. Finally, a new parameter which can express different stress factors and failure mechanisms faced in the field is obtained. This new parameter is combined with failure rate calculations from traditional methods. Therefore, failure rate and return rate predictions are modified. In addition, in this paper two mathematical models of combination of this new parameter with stress based failure rate prediction and failure rate calculated by applying accelerated life tests are given with a real life case study. The results of these two methods and the comparison of the results with the real data are also given.