[BrianG]

From various places and some manufacturers. From what I've read, the issue was mostly with cell mismatch near the end of the cell life. It was also recommended to use matched cells so there is very little to no current flow between the cells. Also, if you do want to parallel, it is best to parallel individual cells instead of creating a series pack and putting it in parallel with another series pack. Any cell mismatches would be amplified in two series packs put in parallel.

I had a bunch of 2500mAh AA cells lying around a while back, so I though I'd try to parallel two for greater capacity. I charged them all and paired them up so there was little mismatch, but even so, there was up to ~0.05v between most pairs. Not a big deal at first glance. However, I made a 7.2v pack and it never seemed to have the capacity of a 7.2v pack with series cells only (of the same type) multuplied by 2. For example, if I was getting 2A for 70 minutes on the series pack, I would expect 2A for 140 minutes on the paralleled pack. It was close, but not quite. I got maybe 90-95% of the expected capacity. I can only guess that the cell mismatches caused the strong cell in each pair to either be undercharged by a little, or overcharged the weaker cell and I lost some capacity. Mybe the test would have come out better using GP cells or something, but I didn't want to waste the time to find out.

So, in a nutshell, it probably won't hurt anything as long as the cells are pretty close to start with, but it looks like you get better results with a single cell of the capacity you want instead of paralleling. Also, at or near the end of the cell life, any mismatches even between individual cells may get worse.