There has been quite a bit of discussion on this.  Will there be enough fees? Will they trend to 0 or to infinity?  Will mining be supported enough?  Its all a big unknown.  If you do some searching on the forums, and looking through the #bitcon-dev IRC logs you should be able to dig up quite a bit.  (there was a discussion with TD/[mike] a couple days ago on this matter after he had a discussion with his coworkers  at Google.  

I believe the OP was asking about mining but in any case.  Current M1 supply of USD is around $1.9 trillion = 190 000 000 000 000 cents. 
M1 of the Euro is around 4.7 trillion = 470 000 000 000 000 cents.  Approximate UK M0 supply is 193 million GBP = 19 300 000 000 pence.  Added together you get 660 019 300 000 000.
Maximum bitcoins in base units is 21000000 0.00000001 = 2 100 000 000 000 000 (you missed a 0). 
Obviously there are way more currencies in the world, but you get the idea.  We have enough.

Note that I use M1 supply here as it represents hard currency not money which is created by financial institutions on top of the currency as those could be created on top of bitcoin and are not bound by the 21 million limit.

http://www.federalreserve.gov/releases/h6/current/h6.htm
https://stats.ecb.europa.eu/stats/download/bsi_t02_03_nsa/bsi_t02_03_nsa/bsi_t02_03_nsa.pdf
http://www.bankofengland.co.uk/statistics/fnc/2011/Mar/TableA.xls

Thanks for the replies guys. I did some more reading on this yesterday. Obviously this scenario is quite unknown, as it will only happen if there is a wide adoption of the currency and if this takes off.

However, my thoughts are that if there are enough people transacting in this currency then it will be in someone's interest to keep the system (ie. required # of nodes) running. And hopefully that person will be able to fund maintenance of the network.

However this is an unknown. Sure the fee system could work but I'm not convinced about that and still a little confused having read over some past forum posts.

Plus isn't there an issue once the network shrinks to a certain size that makes it vulnerable to security attacks?

How secure is this? This is a brand new technology and there are bound to be issues that will not arise until it reaches a certain level of prevalence. There are also bound to be issues with the source code and the technology behind BTC. I hope that I am just being cynical here and that even if any issues exist they will not pose a threat to this technology. However, with open source software, more eyeballs doesn't always mean software is more secure.

I think there will be volunteers but we shouldn't and don't have to count on them. A very cool thing to me is that small scale mining operations have lots of advantages over bigger ones. If you already have a room and a gpu or 4 then your cost is essentially electricity. If you are in a cold area some of that is returned to you in heat anyway, but if you have to buy the building specifically for this purpose and pay technicians etc your cost will be well higher. If people like mining for altruistic reasons it will be even harder for big players to survive.

The main question might just be: do we need lots of miners to support the system at that point?

Optimal case: a system can be established that can agree on a block chain without the use of enormous processing power. Right now, we need expensive calculations to ensure no single person gets all the new coins, but later? There might be a way to keep the system secure and fees low, and we should work hard to find it if it exists. It would make Bitcoin a lot more valuable if there was no compromise to be made between transaction fees and network security.

Bitcoins will always be created. The cap of 21M is just artificial. It is a result of the imposed rate of exchange between the bitcoins attributed to new blocks w.r.t. old blocks. The blocks will still be created at a constant rate, which means that the rate of supply inflation in terms of blocks is ~1/t, while the rate of supply inflation in terms of bitcoins (with the current imposes geometric fall-off) is ~1/t * e^-t.
But people forget that, unlike gold, bitcoins (or for that matter blocks) are not all identical. They have an age. This is bound to have consequences for the relative market price. At some point in time old bitcoins (read blocks) will be values different from new ones.
I have not investigated all the relevant factors that will determine the price(block age) but I doubt that it will remain exactly as prescribed by the current bitcoin client.
Note that, currently there is no way to select exactly which bitcoins you want to send to someone. But there may be very good reasons to add that feature to the bitcoin client. As soon this is done, people may prefer to have old coins or new ones.... I don't know...

Ok, but these blocks will still be assigned to people. What if people start trading them as a separate currency? You see what I mean...

Really? So not all bitcoins have identical value? Am I the only person who read this and thought it a little strange..

Oh, I get it. The first transaction in the block assigns a number of bitcoins to an address that is chosen by the creator of the block. The software is currently configured to reject blocks that assign an incorrect number. That is a  number that is not according to the sequence: 50, 25, 12,5, etc for respectively 4 years each. So you say, that at some point in time the block creator cannot add an extra transaction assigning bitcoins to himself anymore?