Heh... man, I'm glad I could find the post above so that I could make this update. 

I couldn't, for the life of me, get minimal logging to work using INSERT/SELECT on an existing but empty table with a clustered index on it.  The problem was, I was using a Tally function to generate the rows and, to control the number of rows, I was passing a variable to the Tally function, Finally remembering that a variable on something in the FROM or WHERE clause might cause a bad execution plan because the value of the variable isn't known at compile time, I added OPTION (RECOMPILE) to the table population query and, VOILA!, it finally worked as expect.  Built a 7 million row table in under 3 minutes with the clustered index in-place on my laptop.

That little nuance isn't in BOL for minimal logging anywhere.  To be on the safe side, I recommend using OPTION(RECOMPILE) all the time so that you don't have to remember to add it in in certain cases.  The recompile time is trivial compared to the huge savings that minimal logging can give you.