Our summer students were undergrads, typically rising juniors or seniors, who often were outstanding at schoolwork (it was competitive to get a slot), but varied in lab ability. I've had wonderful naturals at labwork, and people who needed to stick to theory to be successful. All in all, I published papers with about half of the students I mentored at National Lab, and gave meaty projects to all of them, allowing them to work as independently as their abilities warranted. There were some people at National Lab who used their students as glorified dishwashers, keeping them very far away from any actual experiments. I think this is a total disservice to the students--they are there to see how science is done and to get some research experience, NOT to do scut work all day long. It also turns excited, interested students off of science, so it always pissed me off. Given the decent levels of support for summer students (it was an REU program), and the fact that we didn't have students the rest of the year, I was happy to take on the responsibility.
This is my first summer at Prodigal U, and I have 3 summer students in the lab. I was not sure I would add summer students right now, because proper mentoring of an undergrad can eat up a lot of time. That said, a major advantage of being at Prodigal U vs. National Lab is the wide availability of students interested in working in the lab for the summer--for pay, for credit, or even as volunteers (I have one of each). They are all really hard workers, and can be used for some tasks that are more appropriate for a short term lab member than a PhD student (such as ultrarisky projects). As a new faculty member, I've found that summer students are a huge help.
I have one summer student working out variant protocols for some important experiments. This frees up my PhD students to get further on taking actual data for publications and their projects. This student really wanted to have their own project, so this is a pretty decent compromise.
I have another student starting out a very, very risky project. If it doesn't work, the student got a lot of interesting lab experience and I didn't screw over a more long term student. If it does work, the summer student will be a co-author on the first publication, and I can put a new PhD student on that project without worrying that the project will fail spectacularly and without usable data.
My last summer student is doing some very important but repetitive measurements. This allows my grad student to work on pushing the envelope on her project, while getting the summer student an awesome poster for the student poster session and authorship on the paper that will almost certainly come out of the data (it is looking really promising right now). This student is really excited to have tons and tons of data to analyze (with a lot of help).
I was really leary of taking students this summer, with it being so important that my grad students make a lot of progress on their research. I screened the students pretty carefully (two I knew from my undergrad class), because I didn't want them to be a distraction in the lab. It has turned out to be a great decision, and I am really glad I did it.