Practical ultrasonic structural health monitoring systems must be able to deal with temperature changes and some signal amplitude/phase drift over time; these issues have been investigated extensively with low-frequency-guided wave systems but much less work has been done on bulk wave systems operating in the megahertz frequency range. Temperature and signal drift compensation have been investigated on a thick copper block specimen instrumented with a lead zirconate titanate disc excited at a centre frequency of 2 MHz, both in the laboratory at ambient temperature and in an environmental chamber over multiple 20°C–70°C temperature cycles. It has been shown that the location-specific temperature compensation scheme originally developed for guided wave inspection significantly out-performs the conventional combined optimum baseline selection and baseline signal stretch method. The test setup was deliberately not optimised, and the signal amplitude and phase were shown to drift with time as the system was temperature cycled in the environmental chamber. It was shown that the ratio of successive back wall reflections at a given temperature was much more stable with time than the amplitude of a single reflection and that this ratio can be used to track changes in the reflection coefficient from the back wall with time. It was also shown that the location-specific temperature compensation method can be used to compensate for changes in the back wall reflection ratio with temperature. Clear changes in back wall reflection ratio were produced by the shadow effect of simulated damage in the form of 1-mm diameter flat-bottomed holes, and the signal-to-noise ratio was such that much smaller defects would be detectable.