I really would implore you not to blow £3000 on an unnecessarily equipped system, but if you insist, here it is;
Seriously though, there is no earthly reason you could have for spending that much money on a system. There is a serious case of diminishing returns here. You just won't notice much difference at all from a £1500 system. You've got quad-core and 8800GTX cards in SLI and 4gig of RAM in this setup with 3 Seagates for a RAID5 array and a Raptor for your OS install, a top-quality Lian-Li server case, and really, what's the point?
Nobody needs 1.5 terabytes of storage unless they're running a server, in which case you don't need twin GTX cards because they're only going to matter for gaming. If it is a server, you should dedicate the £3000 towards a SCSI array and a T3 connection. If it's an archive, use arrays of smaller disks in a data-securing RAID configuration. Why can't you scavenge your old disks anyway? Or are you intending to have 3TB overall?
Matter of fact, who really needs twin GTX cards at all? Superhumans who can perceive the difference between 150 and 200 frames per second? Because otherwise it's a waste. In fact, as I've said elsewhere, I don't see justification in buying a DX10 card at all at the moment; better to buy a mid-high range DX9 card, and keep the saved money towards the inevitably cheaper and more powerful DX10 cards that will come out when DX10 becomes necessary. An X1950XT will serve most any gamer's needs at the minute.
4GB of RAM will only matter if you intend to upgrade to Vista 64. In fact, in anything other than XP Pro 64-bit edition, Windows won't even utilise all 4GB of memory ever. It can't.
And you'd be paying £600 for a quad-core processor, which is only really justified for the hardest-core power users who multitask between various graphics and video encoding apps at once, or again, servers. Lots of people like to delude themselves with the belief that they're power users, when in truth they're just sacrificing their cash at the altar of vanity. A 2.4GHz E6600 costs £400 less, and if overclocked will reach similar speeds to what an overclocked Quadro can obtain. Buy the E6600 and a £35 aftermarket cooler and learn to overclock, and save yourself a mound of cash. It's got 4mb of L2 cache, which will provide the performance boost in video-encoding in comparison with the cheaper E6300 and E6400 Allendales, and the quad core advantage would only ever have become apparent in very, very limited circumstances and very limited ways. It just doesn't warrant another £400.
It's all a case of sitting back and really considering what is fit for purpose rather than being recklessly extravagant. PC components suffer a massive case of diminishing returns. You could, for instance, shell out even more money on this system and buy Dominator-8888 chips at £450 a pair instead of £210 for the one's I've specced you. But the difference evaluated in any real-world benchmarks between the two chips is 0%.
Why do this? Why waste the money you work hard to save? Look at the £1000 specs people are building. These are good machines that will perform nearly as well in the real world as your £3000 machine. If you must, spec £2000 and go for a big SLI rig with a 1kW PSU and 2x8800GTX, if you're an insane gamer. Or throw in a Quadro if you really are a high-end power user/software developer/graphics processor for some multinational. Or tonnes of storage if data-serving is your calling. But save the £1000-2000 difference toward your next build instead. A £3000 system now will lose value immensely and is a far less cost-effective method of futureproofing than moderate, sensible investment coupled with future upgrades, rebuilds or replacement of redundant parts.
Hell, give £1000 to charity instead, save hundreds of lives with clean drinking water or build a school. They'll name a village after you in some parts of the world for that.
I'm not intending this as a flame and it doesn't come from jealousy. If you're asking for the best way to spend £3000 then I feel obliged to point out that a very large portion of that money - at least £1000, maybe £1500 or £2000 depending on your intentions for the system - would be expended wastefully and would be much better reallocated to future investment. However, as has also been said, it's your prerogative to do as you like with your money. As has been excellently surmised, people pay hundreds of thousands for a Veyron that in the real world is still obliged to obey a 70mph speed limit, or thousands on a Rolex that, being a quality mechanical timepiece, will not keep time as well as a cheaper quartz wristwatch. The difference is you can drive that Bugatti around from now 'til Doomsday and people will admire it as a beautiful triumph of engineering, a motoring masterpiece; it will hold value as a classic. The Rolex will always be a Rolex and will be admired for many years to come. This is not true in the world of pc hardware. In 18 months' time a £3000 base unit will be matched in performance by an average new system of the day, and will be effectively obsolete. It won't be worth a fraction of its original price.
It's not quite the same as buying a Bugatti. It more closely analogises to throwing your money in a pond.