In most C or C++ compiler, arithmetic shift are adopted.
With this, negative number shifting may first looks a little out of routine, or at least that's how I felt.
In brief, arithmetic shifting complementing the empty bit with the value of the sign bit.
Like:
a = 1111 1111 1111 1101 (-3)
a >> 1 would be 1111 1111 1111 1110
a << 1 would be 1111 1111 1111 1011
and with a positive number:
a = 0000 0000 0000 0011 (3)
a >> 1 is 1
a << 1 is 6
which should be known to all.
So we see that, arithmetic shift are different from logic shift when dealing with negative numbers, it use the sign bit for complementing.
For more of this I recommend wiki, and this page http://stackoverflow.com/questions/4009885/arithmetic-bit-shift-on-a-signed-integer .