The applicability of the traditional compact expressions for estimating the regional signal-delay times of heterojunction bipolar transistors is examined by means of a comparison with numerical device simulation. Regional partitioning in the numerical simulations is achieved by means of a simple phenomenological scheme. It is shown that the traditional expression for the emitter delay is inadequate due to a neglect of both the stored free charge within the emitter-base, space-charge region, and the stored charge in the quasi-neutral emitter. The importance of properly choosing the effective collection velocity when estimating the base delay is demonstrated by considering the effects of band-gap narrowing, quasi-ballistic transport, and hot-electron injection. It is pointed out that the effects of finite charge in the collector-base, space-charge region cannot be neglected when computing the collector delay.