I would advise you to consult during the project with someone who has experience with this. There are a number of things to take into account
1) Ethics. You will need new approval
2) Training. In order to teach kids to stay still in the scanner, most groups go through practice sessions with them in a mock-up scanner. Usually with some sort of customized movement feedback program.
3) Stimulation. You cannot have children perform all tasks as easily as adults. For example, you may need to use metaphors to explain to them why they need to "fixate on the white cross"
4) Rewards. Will you reward the participant, their parents, both?
5) Setup. Will the child fit properly in your head-coil? Same with mirrors, headphones, etc.
And other things come to mind. In many case you have to deal not just with your participant's anxiety (if any, many kids are very cool about it), but also their parents. Take extra time and keep them busy during breaks etc. On the whole, it is a lot more involved. I probably mentioned only a subset of the challenges you'll have to deal with. Let me know if you have any further questions.
Good points from Jasper. The other obvious thing that came to my mind is related more to the analysis than carrying out the session - particularly coregistration/normalisation. Depending on what age the kids are (and therefore how big their brains are) you'll probably find that using adult template brains for coregistration and normalisation is sub-optimal. You'd probably be much better off using one of the paediatric brain templates from the MNI:
Another issue is head movements. From my experience scanning children and adolescents it is best to constrain your scanning sessions to short 5 min blocks, for while adults can remain stationary for up to 10-20 min children tend to move and fidget so you are best off with short runs. i also suggest counterbalancing the runs such that if you lose one this does not ruin your whole experiment.