The last 30 years have witnessed an enormous growth in fixed-income markets. How long-term fixed-income strategies should be implemented for the welfare of investors has become a major concern of bond managers. This study makes use of stochastic optimal control to formulate a multi-period portfolio selection model and implements it using backward recursion algorithm to find numerically the optimal allocation of wealth between long- and short-term bonds for an investor with power utility and an investment horizon of 10 years. By way of a technical manipulation, this study uses the fact that the long rate and the spread (difference between long rate and short rate) are uncorrelated to simplify model formulation and parameter estimation. The results show that an investor would increase his/her holding of short-term bond if his/her investment horizon becomes shorter or if he/she is more risk averse, or both.