Reverse voltage at rundown, for one. I assume you're looking for 12 volts (your cell count gives you 10 cells), so they'd be series connected (also, if the "3 to D" were parallel models, why not just go with 4 D cells?).
When you series connect cells of differing capacities, an interesting thing happens as you get close to fully discharged (note: this is why you can be more agressive with discharging single cells than packs - there's always a bit of difference in capacity between cells in a pack). Naturally, the lowest capacity cell will discharge first. If the others still have "juice", they'll put a reverse voltage across the "weak sister" - one of the quickest ways to destroy a cell.
Possible solutions:
1. 3x "3 to D" series, 1x "3 to D" parallel, with only 1 cell in the parallel unit. DO NOT put dummy cells into the empty spaces - if you need to fill them for mechanical reasons (i.e. to keep the "lonesome" cell from shifting position), use nonconductive spacers.
2. 4x "3 to D" series, with one of the units holding a "live" cell and 2 dummies.
3. 3x "3 to D" series, 1x adapter sleeves ("AA to C", and "C to D"). I believe there's a thread somewhere around here on where to find those.
All of these will give you 10 cells of the same (nominal) capacity in parallel, but you still have to watch out for random differences (i.e. treat it as any other series-connected pack so far as not draining it all the way is concerned).
