I have figured out that assigning transient recharge to two non-contiguous polygons seems to result in the second polygon not actually being assigned the recharge values. For whatever reason, probably the way in which I am trying to enter the data, I can work around this by creating two seperate polygons and applying the recharge seperately.
I still have a slight problem with long term heads in a transient model. The model has identical total annual recharge as the steady state mode, except it is applied in a seasonal pattern i.e. high winter rainfall and low summer. For some reason, the model drops below the st.st heads by about 0.25m during the first few years, then as the model continues (20+ years) the heads tend towards values ~0.25m above the st.st. values. I cannot think of an explanation for this!