Negative bias temperature instability (NBTI) prediction relies on a reliable extraction of power exponents from its kinetics. When measured by fast pulse technique, however, the kinetics does not follow a power law. This paper reviews the recent progresses on how to restore the power law, based on the As-grown-Generation (AG) model. For nanometer sized devices, NBTI is different for different devices, inducing a time-dependent variation. The new technique proposed for characterizing this Time-dependent Variation accounting for within-a-device-Fluctuation (TVF) will be reviewed.