Abstract An approach for assessing reliability data provided by compound semiconductor foundries is presented. This approach combines a thorough review of the foundry results with independent accelerated life tests to confirm these results. Examples are provided of cases where these independent test results did not adequately match the results provided by the foundry. Two factors are shown to be especially important: consistent temperature estimates by the foundry and the customer, and an understanding of the impact of bias conditions on the acceleration of failure mechanisms in FET and HBT technologies.