GaAs MMICs (Monolithic Microwave Integrated Circuits) reliability is a critical part of the overall reliability of the thermal solution in semiconductor devices. With MMICs reliability improved, GaAs MMICs failure rates will reach levels which are impractical to measure with conventional methods in the near future. This letter proposes a methodology to predict the GaAs MMICs reliability by combining empirical and statistical methods based on zero-failure GaAs MMICs life testing data. Besides, we investigate the effect of accelerated factors on MMICs degradation and make a comparison between the Weibull and lognormal distributions. The method has been used in the reliability evaluation of GaAs MMICs successfully.