Almost forgot: Assuming you use VB (which is a bad assumption) you can use 4 * Atn(1) to get VB's approximation.
In ANSI C, you can use atan() which is in math.h.
In Intel's FPU there is an instruction, FLDPI (Opcode: D9 EB), which pushes the FPU's definition of Pi into the floating point stack.