Mil-std-217f

Part failure models in MIL-HDBK-217 are based on part data from numerous sources. Complete models are hardly ever developed under a single study contract nor are the failure data from a single source. For example, all MIL-HDBK-217 environmental factors were developed under separate study efforts from the part failure models. Therefore, because of the fragmented nature of the data and the fact that it is often necessary to interpolate or extrapolate from many available data sources when developing new models, no statistical confidence intervals should be associated with the overall model results.
Sometimes, you may hear 60% confidence mentioned. However, this applies to analyzing data to develop the underlying base failure rates when no failures have occurred. For example, to calculate MTBF when you have accumulated some experience time, you cannot divide time by zero failures, so you either assume one failure (and “penalize” yourself, so to speak), or use the chi-square distribution, assuming time truncation with zero failures. The later approach is commonly used and will provide a higher MTBF estimate than the former. However, this has sometimes been construed to indicate that “MIL-HDBK-217 predicts to a 60% confidence level,” which is incorrect because of the other factors noted above.

I assume it is 50%.
The reliabilityat other confidence levels can be determined using chi-square statistics – see Reliability for technology, Engineering, and management by Paul Kales and pub Prentice Hall

Sir,
Thank you very much for your kind assistance.