A method for determining whether a first group of a product, a component or a system in reliability life testing has longer lifetime than a second group. This method is non-parametric and free from a pre-assumption of statistical distributions and can be applied to all kinds of data and distributions. Errors from goodness-of-fit of distribution fitting and parameter estimations are thus eliminated. After pre-check on bimodal, early failures, and the failure mechanisms, the method employs numerical solutions with good accuracy by the nonparametric approach. The data under consideration can be censored, interval or bimodal, and not limited to simple cases of complete type. The method can be used to determine multiplicities of reliability tests for all product types and at all levels. Based on a comparability index derived from integrating the weighted difference between the reliability functions of the two groups under comparison. Several indices are proposed for effectiveness of reliability comparability.