Scalable Montgomery multiplication methods and apparatus are provided that are reconfigurable to perform Montgomery multiplication on operands having arbitrary data precision. The methods perform Montgomery multiplication by combining bit-wise and word-wise operations and exhibit pipelined and parallel operation. Apparatus include a control unit that directs bits of an operand to processing elements that receive words of a second operand and a modulus, and produce intermediate values of a Montgomery product. After an intermediate value of a word of a Montgomery product is obtained in a first processing element based on a selected bit of the first operand, the intermediate value is directed to a second processing element and is updated based on another selected bit of the first operand.