Reference/recommendation letters from undergrad?

[Butterfly_effect]

Fellowship season is upon us, but as a second year student, I only probably have one potential letter writer at my grad institution (my PI). Most fellowships require 3-4 letters. Is it a horrible thing to have 2 of the 3 letters be from undergrad? For me personally, these would be from two different institutions (one REU, one my undergrad college) with PIs that worked with me and I ended up being on a paper in each lab. I just don't think anyone from my grad institution would know me as well. There have been professors whose classes I've done well in, but they can't really speak to my research ability. 

[fuzzylogician]

I would suggest talking to your current advisor about this. In principle I would think a 2/2 distribution would look much better, but this really comes down to the content of the letters on the one hand and how the selection committee is likely to use the letters on the other. Your advisor will know more about the content of the letters and presumably also about the selection process for this and/or similar fellowships. To me, a reasonable compromise would be 3 strong letters: advisor + 2 from other research experiences, and a weaker letter from someone else at your current program. 

This might be a good time to start thinking about cultivating relationships with other people in your program beside your advisor, because this problem is certain to come up again.