I would look into understanding the physics of the problem if you are trying to get a realistic simulation. I'm sure you can google just as well as I can, so I'll just.....
*cough*
http://www.myphysicslab.com/pendulum1.html
*sneeze*
http://www.physicsclassroom.com/mmedia/energy/pe.html
*gulp*
http://scienceworld.wolfram.com/physics/Pendulum.html

If you are going for ultimate accuracy, I would suggest that you do not use a timestep method that may seem fairly intuitive. I.e: You could calculate the force, then apply the force over a timestep of say 1/60th of a second. The problem with this method is that the pendulum moves in a linear fashion during the timestep... thus, if your timestep goes up, the pendulum gets more and more "off course".

For the best accuracy, I would simply create an equation based on the length of the pendulum's "rope". Then plug the time into the equation each frame and recalculate the position each frame.

Hope that helps.