I would recommend you to do your homology modeling using multiple templates.
See the tutorial on homology modeling using multiple templates here. Here is an interesting manuscript that utilized such approach for modeling, please see section "2.1 Selection of templates and input preparation" for details.
There is more than one possible way to go from here, which one is the best depends on your protein and the homologies you find.
- Model the whole protein in one go (HHpred as Jan suggested, SwissModel)
- Join your two models together and perform energy minimization
Modeling the whole protein in one go is of course the desirable option if you can find a homologous protein that covers the entire sequence. In any case, validation of the structure you find is important. Energy minimization as means to orient the side chains properly may also be needed when using one of the homology modeling suites, but I guess you already know that.
As michael Schubert said, modeling the whole protein would be the best. If you have two homologous structures with different length, you can still use both of them at the time to make a model. However, I have experience of doing that in Discover Studio, a commercial modeling software. If you are using that program, you can refer its tutorials which is quite detail. And yes, as long as you have similar template structures, you can use more than one template to make homology model.
I-TASSER would be another alternative for your solution. Just try that too because you will find out how this program modeled N-terminal that you had modeled separately and also compare the models as a whole.
First you need to check if you can find any PDB template that has been solved with both N-terminus and C-terminus. You would use that template for orienting the termini relative to each other. So take the full length protein sequence, run HHPred and check if you get any full length hit.
Once you do it, let me know what you get, you can paste HHPred link to results if you want, I help you with further steps.
Also tell us if the relative orientation of termini is important for you at all.