If a lot of miners leave to a different fork, you're into 51%ing trouble. A lot of it (especially with ASICs around).

The firm I work for, which has made substantial investment into Bitcoin, has already thought quite a bit about this.  This is what we're doing and thinking:

* While this block size issue remains such a big uncertainty, we have drastically slowed our pace of investment and we've shelved some start-ups that were already in progress.  We're not stopping or backing out, but we're proceeding much slower and more cautiously until a clearer resolution to this problem appears.

* If we approach the 1MB limit and a solution does not appear forthcoming, we'll cease all new investment.

* If we pass the 1MB limit without a solution, seeing even the slightest hint that Bitcoin's competitive advantages over the conventional banking and payment systems are being eroded due to an inability to scale, we'll dump all our bitcoin assets and holdings.

* If a controversial solution is proposed, with fierce arguments on multiple sides, we will follow Gavin's fork, even if it's not ideal, on all our sites that accept BTC.

* If the fork hits, and there's even the slightest uncertainty as to which will survive, we will immediately dump all our bitcoin assets and holdings.  We will remove Bitcoin from all of our sites that accept multiple payment methods -- at least while we wait to see how things play out.

* If one fork kills off the other, we'll adopt that one and go back to business as usual.

* If both forks survive, with both being widely accepted (for example if Mt.Gox begins accepting Bitcoin-A and Bitcoin-B), we'll accept neither, dump everything, and write off blockchain based currencies and too risky and too unstable.

What we really hope to see is a nice, smooth transition to a system that scales, which very large majority of the network, including major service providers, agree to.