Deciding to mine is an important matter, also because this is a contribution to the network stability. How much can one decide to risk in this anyway?. I decided to share my analysis of the situation here (sorry, I don't know if such a self linking is ok with the TOU):

http://laurent.henocque.com/post/2013/09/16/On-Bitcoin-Mining-and-the-vallue-of-Bitcoin

My conclusion is that to account for risks, the value of a BTC is grossly fourfold underestimated. The reason is that to mine, one must expect obtaining more BTC (not $) than by buying them now.

BTC break even can be theoretically expected using mining hardware paid today (assuming 139$/BTC - but bitstamp rates make it worse) at:
    16 dollars/GH if you expect mid october delivery
    11 dollars/GH if you expect early november delivery
    8 dollars/GH if you expect mid november delivery
    6 dollars/GH if you expect early december delivery
    4 dollars/GH if you expect mid december delivery
    3 dollars/GH if you expect early january 2014 delivery
    ...

However, if you look at the table, the correct figures are in BTC
 
To be on a 'safe' side should mean to at least anticipate a one month late delivery, and to expect doubling the BTC. So 3$ / Gh paid today for a delivery in january is not good. I would not accept to pay more than 0.7 $ for this.

More badly, a KNCMiner Saturn expected 'sometime in October, or November' will never mine the BTC it cost (from June to now).

I may be wrong indeed...

(edit - typos)