I am writing a paper with a group of people, and we would like to cite preliminary results from another work, which one of the coauthors is involved along with yet other people. The preliminary results have not been published or even written up. Should we cite this as "Personal communication" and put the names of all members of the other group (which overlaps with the current group), or are there better ways to do it?
I hate to disappoint you, but you would be referencing the material which is impossible to get. The reviewers do see it similary and complain, and they are right in doing so. Personally, if there is any way whatsoever to convince the other group to write up the preliminary results and put them on a web page in time, do it. See point 7 in most harmful heristic. Probably, the best you could do now is saying "to be published in [SB]" where the reference is a half-hearted entry along the following lines:
[SB] John Smith and Joe Blotz, The Importance of Being Published, 2017, in preparation.
Adapt the title, of course.
My experience with "Personal communication" citations has been that editors require you to present something from the person you're citing saying that yes, they really did communicate that to you. I've only done this a couple of times, so I can't give a wide range of conditions, but we've only done this when we really have to. For example, in one case we cited a paper that had an important typo that the author was in the process of correcting, but it was slow and going to be well after our paper was out; the author told us this at a conference, we noted the correct information as (Personal communication), and included an email from the author explaining the issue.
I would be reluctant to include anything more substantive than that as "Personal communication". I've occasionally reviewed papers that offer some major part of their story as "Personal communication", and rejected the papers specifically for that reason; if the supporting evidence hasn't been and can't be peer reviewed, then the whole paper is unsupported and needs to wait until the other paper is at least accepted.
As Leon Meier notes, "if there is any way whatsoever to convince the other group to write up the preliminary results and put them on a web page in time, do it." During review you also have an alternative: they could make their results available only to reviewers, e.g., by submitting along with your manuscript (if this is supported) or by uploading to Google Docs, Drop Box, or similar, and removing after reviewing is complete.