How page ranges work
One notation covers extracting, splitting and deleting: a short list of numbers and dashes that says which pages you mean.
How do I write a page range?
As a comma-separated list where a dash means a run. “1-3,7,10-12” means pages one, two, three, seven, ten, eleven and twelve. A dash with nothing after it, as in “10-”, means from ten to the end, and a bare number is that page alone.
Order matters less than people expect, and then suddenly matters a great deal. “5,1” removes the same two pages as “1,5”, but when you are extracting into a new document it decides which page comes first, so a reordering job is a range deliberately written out of sequence.
Which page is page one?
The first sheet in the file, always, whatever is printed on it. A report whose cover carries no number and whose body restarts at 1 still has its cover as page one here, because a range refers to positions in the file rather than to printed folios.
This is the single most common cause of an off-by-one result. When the numbers printed on the page and the page count in your viewer disagree, trust the viewer: that is what the range is counting.
What happens to the pages I do not select?
It depends which job you asked for. Extracting copies the selected pages into a new document; deleting produces a copy without them; splitting produces several documents that between them hold every page exactly once.
None of the three touches the file you started with. Every tool here writes a new document and hands it to you, so a range typed wrong costs you a download and nothing more.
What travels with the pages, and what does not?
The page content, its annotations and its form fields come along. Bookmarks pointing at a page you kept are carried over; bookmarks pointing at a page you removed are dropped, because a bookmark to a page that no longer exists is a broken link in every reader.
Internal links behave the same way. Document-level things stay behind: the metadata, the attachments and the outline structure belong to the original, so a five-page extract begins with a clean set of its own.
Where to go from here
Extract pages takes a range and gives you those pages as a new file, split cuts one document into several at the boundaries you set, and delete pages returns the same document without them.
Merging and what happens to bookmarks covers the other half of the outline question, and why editing PDF text is limited explains why moving pages is easy and moving words is not.