I would say work with an established, characterized line; do NOT set one up yourself.
I took a risky transgenic mouse project for my thesis, since I had three publications from my undergrad work and didn't feel the immediate need to publish. I spent three years trying to make the mice, screening, western blotting, recloning a new construct, waiting for offspring, etc. In all I screened over 600 mice by PCR and/or southern, plus I had a tissue specific transgene that required sac'ing the animal to see if protein was present. In the end, I tried 4 different constructs with 7 sets total of microinjection, and 0/600+ had expression.
Lucky for me I have had an alternative pathway using transplantation to test the effect of my transgene is a slightly different setting. I have also had failures here, but lucky for me I have had a lot of success lately, and all of my pubs will come from the last 1.5 years of my grad work.
Keep in mind that my committee said that I had enough data for a thesis as of last summer, but my department requires 3 first authored papers to get the PhD. Thankfully, I have two that are ready to be submitted, and a third set of experiments based on these data are underway and should produce another paper for me.
If you have any specific q's about working with Tg mice, give me a PM.
Treg