A method for operating a Viterbi decoder uses few data move operations to improve efficiency. The Viterbi decoder predicts a state in which the convolution encoder might have operated while generating a convolutionally encoded data stream. The Viterbi decoder maintains a first set of states and based on the received convolutionally encoded data stream, predicts second and third sets of states. The Viterbi decoder then calculates first and second sets of decision bits based on the transitions to the second and third sets of states. Path metric values associated with the third set of states are stored in a memory buffer. Thereafter, during trace-back, the Viterbi decoder extracts first and second decoded bits from first and second sets of decision bits respectively.