There are many different kinds of moderators that can be used.
The first thing to consider is probably that the moderator has the right physical characteristics to be compatible with the system. Graphite works will in MSRs and VHTR because it can withstand high temperatures and is compatible with the coolant (molten salt and gas). For example, the pressure of water and D2O would be very high at these temperatures.
The second thing to look at is probably cost. Graphite is relatively cheap compared to something like Beryllium.
Finally, if you just want the most effective moderator, there are three things to look at. (1) You want a small atomic mass so that the neutron loses the most energy per collision. In this regard, hydrogen is the best because it has the smallest atomic mass. (2) You want a moderator with a high macroscopic scattering cross section. You can have a high energy loss per collision, but it won't do any good if the scattering cross section is low. (3) Finally, you want a moderator with a low macroscopic absorption cross section. In this regard, deuterium has the lowest absorption. If you put these three things together, you can define a "moderating ratio" for different types of moderators. A list of moderating ratios is shown below:
* H2O 71
* D2O 5670
* He 83
* Be 143
* C 129
* Na-23 0.725
* U-238 0.0092
From this list, D2O is the most effective moderator. However, D2O doesn't have the right physical characteristics to work in a high temperature reactor.
Carbon/graphite has the third highest moderating ratio from the list above, it works at high temperatures, and is relatively inexpensive.
(See most reactor physics textbooks, such as Duderstadt and Hamilton, for a more detailed explanation.)