If you have many links, that solution is of course not the best one. I think,
as I said, that the pdf's may lie in directories not following the starting
one.

Consider this exemple. Suppose you start at url
<http://domain/.../some-directory/page-of-hundreds-links.html> (i.e. the
starting entry "Web addresses (url)" in Httrack), and suppose that you put in
the Httrack "Options>Limits>Maximum link scannin depth" the value 2 (i.e. you
want to download the starting url and all the pdf' links lying there in that
url). Suppose finally that the paths of the pdf's lie in a directory uphill
relatively your starting one (=
<http://domain/.../some-directory/page-of-hundreds-links.html>), e.g.
<http://domain/pdf-folder/>, or that they lie elsewhere, in another domain, e.g.
<http://another-domain/pdf-folder/>. 

At this point, if in your starting url there are many other links pointing at
resources you are not interested in, you will put in "Options>Scan rules"
something as -* +http://domain/pdf-folder/*.pdf (or
+http://another-domain/pdf-folder/*.pdf if the links point out of the starting
domain), in order to filter the links you are not interested in. Otherwise, if
the initial html page contains only links pointing to the pdf files you are
interested in, you will not put anything in "Options>Scan rules" because there
will be nothing to filter.

Under these hypothesis, with the default settings in
"Options>Experts-only>Travel mode" and "Global travel mode", you will not
succeed in your download. You can bypass this limitation, by selecting the
alternative values in those options, i.e. "Travel mode = Can both gi up &
down", and "Global travel mode = Go everywhere on the web".