A different analysis
If we build a Fighter20 who uses the two-bladed sword as his weapon of choice, we should be able to get a good analysis of the relative strengths and weaknesses of two-weapon versus two-handed fighting, since this same character is capable of both. Now, it's worth noting that such a character will necessarily have spent all the money and feats on buffing up his two-weapon style, which the two-handed specialist will have spent those same feats on something else, so it's not a perfect comparison. Still, it's probably not too bad.
So, here's the test character I propose to use:
Fighter 20.
Str 26 (15 start, +5 level, +6 item)
Dex 19 (13 start, +6 item)
Feats: Exotic Weapon Proficiency, Weapon Focus, Weapon Spec, Greater Weapon Focus, Greater Weapon Spec, Improved Critical (two-bladed sword), Power Attack, Two-weapon Fighting, Improved Two-weapon Fighting, Greater Two-weapon Fighting
Weapon of Choice: Two-bladed sword. (+5 Speed Shock/+5 Flaming Frost) (I assumed the character would spend about 33% of his starting funds on his primary weapon, and this weapon reflects that. I think the exact value of the weapon is 260,700gp.)
Two-handed use: +35/+35/+30/+25/+20 on attack rolls, 24 average damage.
Two-weapon Use: +33/+33/+28/+23/+18 on attack rolls, 20 average damage and +33/+28/+23 on attack rolls, 20.5 average damage.
Assuming the character knows the AC of his opponent (which is a poor assumption, I know), and so can optimise his power attack, we get the following:
AC THF PA DMG TWF PA DMG DMG WINNER
10 20 344.82 15 292.65 344.82 THF
11 20 366.61 14 287.25 366.61 THF
12 20 328.4 13 281.85 328.4 THF
13 19 318.8 12 276.45 318.8 THF
14 18 309.2 11 271.05 309.2 THF
15 17 299.6 10 265.65 299.6 THF
16 16 290 9 260.25 290 THF
17 15 280.4 8 254.85 280.4 THF
18 14 270.8 8 249.56 270.8 THF
19 13 261.2 7 244.28 261.2 THF
20 12 251.6 7 239.11 251.6 THF
21 12 242.165 6 233.95 242.165 THF
22 11 232.925 6 228.9 232.925 THF
23 11 223.85 5 223.86 223.86 TWF
24 10 214.97 5 218.93 218.93 TWF
25 10 206.255 4 214.01 214.01 TWF
26 9 197.735 4 209.2 209.2 TWF
27 9 189.38 3 204.4 204.4 TWF
28 8 181.22 2 199.6 199.6 TWF
29 8 173.225 1 194.8 194.8 TWF
30 7 165.425 0 190 190 TWF
31 6 157.625 0 182.805 182.805 TWF
32 5 149.825 0 175.61 175.61 TWF
33 4 142.025 0 168.415 168.415 TWF
34 3 134.225 0 161.22 161.22 TWF
35 2 126.425 0 154.025 154.025 TWF
36 1 118.625 0 142.3388 142.3388 TWF
37 0 110.825 0 130.76 130.76 TWF
38 0 102.1088 0 119.2888 119.2888 TWF
39 0 93.52 0 109.2288 109.2288 TWF
40 0 85.05875 0 99.16875 99.16875 TWF
41 0 78.23875 0 88.81625 88.81625 TWF
42 0 71.41875 0 78.65875 78.65875 TWF
43 0 64.4075 0 68.69625 68.69625 TWF
44 0 57.52375 0 61.50125 61.50125 TWF
45 0 50.7675 0 54.30625 54.30625 TWF
46 0 45.6525 0 46.81875 46.81875 TWF
47 0 40.5375 0 39.52625 40.5375 THF
48 0 35.23125 0 32.42875 35.23125 THF
49 0 30.0525 0 28.09875 30.0525 THF
50 0 25.00125 0 23.76875 25.00125 THF
51 0 21.59125 0 18.985 21.59125 THF
52 0 18.18125 20 15.575 18.18125 THF
53 20 18.06875 20 15.575 18.06875 THF
54 20 18.06875 20 15.575 18.06875 THF
55+ 20 18.06875 20 15.575 18.06875 THF
(I hope that's legible. Also, I hope it's correct.

)
As can be seen from the table, the 2-handed Power Attack mode of combat is superior for the extremely low and extremely high AC values. For the mid-range values, two-weapon combat is superior. Since you would expect those mid-range AC values to be the most commonly encountered, this suggests that two-weapon fighting is the way to go.
However, I would point out again that this is only one example. A character optimised for two-handed combat (using a greatsword) won't have spent all those feats on two-weapon combat, and won't have spent as much money on boosting his primary weapon. How much of a difference that makes is uncertain.
Edit: I have changed the numbers in the table above following a number of corrections. Basically, the damage values for the various attacks were incorrect, and I'd failed to take into account the fact that extra d6's for energy damage don't double on a critical hit. (And thanks to Frisbeet for pointing out the errors, and for his spreadsheet, which allowed me to check the revised numbers - they might still be wrong, but if so it's due to a copying error on my part.)