This International Standard specifies procedures to compare two observed – failure rates; – failure intensities; – rates/intensities of relevant events. The procedures are used to determine whether an apparent difference between the two sets of observations can be considered statistically significant. It is assumed that the time intervals to/between the failures (events) are independent and identically exponentially distributed during the observation period (that is, the accumulated relevant test time). NOTE – This assumption implies that the failure rate/intensity is constant. It is furthermore assumed that there are technical or other reasons to believe that a difference (either an improvement or deterioration) might exist between the observed reliability characteristic of the two sets of items under comparison. Some examples of typical applications are described in 5.4. The methods are designed as hypothesis tests which state, with a specified risk (the significance level), whether the two series of observations belong to the same population or the same process, that is they have the same true mean value. NOTE – Failure rate, which is relevant to non-repaired items, is associated with a distribution of times to failure. Failure intensity, which is relevant to repaired items only, is associated with a point process describing a sequence of events, for example times between failures on a time axis. The procedures are not restricted to comparison of failure rate/intensity, but can be applied to observations of two series of any relevant events, provided the above assumptions are valid. NOTE – The two series of observations may be of items from the same population, or the same item under different conditions (for example environment and load), or just comparable series of events (for example car accidents on a road). Numerical methods and a graphical procedure are prescribed. The observation periods relevant to the two series do not need to be equal, but if they are, the methods are very simple.