First, I would second @AlanGilbertson in go and have a look at Chicago. These thesis templates might seem dusty and a tad boring, but there are some good arguments for them:
International standards. Should your thesis ever be of interest other places, a standard might be useful.
It will save you time in the long run, as soon as you have your document setup. Though that can be fiddly.
These standards might have taken into account things you have not thought about yet. Nothing is more annoying than after 123 pages realising that you have to go through every page to do something tedious because you did not think of it.
Flexibility
Whatever solution you come up with, you should make absolutely sure that your document is as flexible as possible. With some luck your eminent thesis might be the basis for articles, presentations, keynotes etc; so do not use archaic software or bizarre or uncommon fonts. Most journals will ask you to provide text according to their standards, so do not lock your document into something that would be a hassle to reformat down the line.
Readability, simplicity
The absolute most important thing is to make sure it is as easy and comfortable to read as possible. This to ensure that you do not annoy the readers. You want to make it as smooth for the sensors as possible, and you want them to focus on the content, not be annoyed at the packaging. For you to shine in a thesis, you will have to do it through words and media (images, video, graphs, diagrams, audio...), not through creative expression in layout. You must make navigation and referencing as simple as possible.
Of course, there is the possibility that your thesis breaks the boundaries of a standard thesis, and then that needs to be considered too, but the standards will contain pretty much everything. If your thesis is - for example - heavily dependant on links to web, software, prototypes, consider making your own way of highlighting these important links. Do not let these essential bits drown, just because the Chicago style have not accounted for it.
Pagination
In academia, remember: no one can – or will! – reference something that does not have a page number. If there is content you or others need to refer to; it must have at the very least a page number of some sort.
University standard page
Your university may not have a standard for the whole shebang, but most will have a "required-box" or even a page of some sort, with uni name, thesis title, copyrights, contact info, dates, advisors, stakeholders and keywords. Where to put this is a hassle. I say get it over with as soon as possible.
Personal opinion..
So, to my personal opinion. This depends on how complex your thesis will be, but I belong to the "keep it as simple as possible but not one bit simpler".
I partly follow the Chicago model, but I move content around a little, and I simplify a great deal. I go with roman numerals all the way to the end of the table of content. I also add in white pages to make sure it works well when printed. Do not be afraid of empty left pages. Use them when sensible.
Personally, I would stick ALL lists of illustrations, tables etc in an appendix. I see no reason why the reader should plough through lots of lists of stuff s/he has not even seen. Use Image captions, image numbers for reference in appendix.
Here is a basic layout of what I prefer:


General hints:
- Spend some time formatting the table of contents so that it looks
good and are immediately easy to navigate.
- 1.5 line-space, font, Times, Times New Roman or very similar.
- Headings: here you have more freedom, but make them clear, simple,
readable with good indents and size differences.
- Make a footer with name of thesis and your name. Make it subtle.
Light gray is good.
- Use appendices
- Use sensible referencing, such as Chicago or Harvard (stay away from the stupid IT-idea of numbers in brackets).
- Make sure your footnotes are well made: they are not AS important as
the main content, you can tone then down a little so not to interfere
with the main stuff.