Mark Byers created a wonderful Shortest Sudoku Solver page, showing shortest Sudoku solvers in many different languages, but the old link seems to be broken now. I remember being amused at the time to notice that several different folks whittled the Perl solution from 188 to 187 to 185 to 184 ... and then along came Ton Hospel and chopped it straight down to around 120! It's sort of deflating when thospel does that to you. :-)

Update: original link courtesy of the Wayback Machine (Aug 2007). The original shortest was based on a version by "Eccles & Toad":

use integer;@A=split//,<>;sub R{for$i(0..80){next if$A[$i];my%t=map{$_
+/9==$i/9||$_%9==$i%9||$_/27==$i/27&&$_%9/3==$i%9/3?$A[$_]:0=>1}0..80;
+R($A[$i]=$_)for grep{!$t{$_}}1..9;return$A[$i]=0}die@A}R
[download]

Here is the audit trail of shortenings:

 * Original Eccles & Toad solution (slightly modified)
 * Mark Byers reduced it to 187 bytes
 * Gordon McCreight reduced it to 186 bytes
 * Pablo Carbonell reduced it to 181 bytes
 * Simon Stroh changed @A=split//,<> to $/=\1;@A=<> to reduce the prog
+ram to 179 bytes
 * Mitsuru Kariya changed @A[map{ ... }]  to map@A[ ... ] to reduce th
+e program to 178 bytes
 * Ton Hospel shortened the program to 121 bytes
[download]

Ton Hospel's 121 stroke solver:

$_=$`.$_.$'.<>;split//;${/[@_[map{$i-($i="@-")%9+$_,9*$_+$i%9,9*$_%26+
+$i-$i%27+$i%9-$i%3}0..8]]/o||do$0}for/0/||print..9
[download]

which can be reduced to 120 bytes, at the cost of using large amounts of memory:

$_=$`.$_.$'.<>;split//;map{/[@_[map{$i-($i="@-")%9+$_,9*$_+$i%9,9*$_%2
+6+$i-$i%27+$i%9-$i%3}0..8]]/o||do$0}/0/||print..9
[download]