The Sloan Kettering web site gives some references which explain the
research behind their nomogram.   I downloaded one of the articles

Postoperative nomogram predicting the 10-year probability of prostate
cancer recurrence after radical prostatectomy.
Stephenson AJ,Scardino PT,Eastham JA,Bianco FJ Jr,Dotan ZA,DiBlasio
CJ,Reuther A,Klein EA,Kattan MW.
J Clin Oncol, 2005 Oct 1;23(28):7005-12

The article has a curve showing the percent of cases which have remained
recurrence free as a function of time up to ten years.  (The nomogram at
the SK website only goes to 7 years.)   The curve decreases, but it
appears to do so at a declining rate.   This is in distinction to what
ron reported about the results in Walsh's paper where the decline
appeared to be roughly linear.   They also state that it is known that
the recurrence rate does decline rapidly in the first few years, and
they refer to a 1997 paper by Scardino and others which described a
study of 611 patients  and addressed that question.  The graph in the
1997 paper shows that the probability of recurrence in any given year
declines precipitously for the first 2 1/2 years, rises slighly for the
next two years and then declines to a low value at year 6 1/2, after
which it remains constant.  But remember there is a lot of noise in such
estimates, and the figures were for all kinds of cases, so don't read
too much into those figures.  Also, this doesn't seem to be consistent
with other studies such as that of Walsh which appear to show a linear
relation between the percent not recurring and time.  I will try to find
other studies which might bear on this matter.

The nomogram based on this research is at the end of the article.  It
shows that a typical Gleason 6, moderate PSA, with no capsular
penetration and negative margins has a likelihood of recurring in ten
years from surgery of at most 1-2 percent.  That is quite consistent
with Walsh's results.  For a case like mine, which was Gleason 7 but
otherwise similar, the likelihood of recurrence was higher, but not by
too much.

The major contribution to the nomogram estimate is actually the
presurgical PSA.  For example, I got 13 points from having a Gleason
7=3+4, but I got 77 points from having a PSA between 4 and 5.

The nomogram also adds some additional information which could be
confusing.   It will tell you, for example, if you have been recurrence
free for 5 years what the likelihood is that you will still be
recurrence free at 10 years.   This reflects the fact that the closer
you get to 10 years without recurrence, the higher the likelihood that
you will still be okay at ten years.  I don't really see the point of
this.  What I would be more interested in today is what would be true
ten years from the present time, not from my original surgery.
Certainly, 8 years from srugery, the 10 year anniversary might have some
symbolic signficance but would not otherwise be specially interesting.

They do say in the paper that once you reach 10 years, the likelihood of
recurrence after that is pretty small, but aside from some rough
estimates they don't go into it in detail.

The nomogram also allows you to enter information on the year of surgery
ranging from 1996 to 2004.   It appears that they found the more
recently you have been treated the less likely you are to recur at any
given time after surgery.   It is not much of an effect, but it still
surprised me that they found there was a measurable improvement in
outcomes over this period.

Finally, their main criterion for recurrence is a PSA > 0.4 following by
a still higher PSA at a later date.   If recurrence were based on a
lower starting value, it would presumably show a higher recurrence rate
at any given time after surgery.  But my guess is that it wouldn't make
 a lot of difference because the important characteristic of recurrence
is that the PSA continues to climb, not that it reaches some specified
threshhold.   Still those defining recurrence differntly might come up
with somewhat less optimistic figures because of that.