Summary: In this study, we investigate a general continuous-time portfolio selection model with loss aversion in an incomplete market where the number of stocks is strictly less than the dimension of the underlying Brownian motion. The investor’s preference facing market risks is defined by an S-shaped value function. By transforming the market into a complete one, we solve the optimal terminal wealth and the optimal wealth-portfolio pair of agents using the martingale approach and a replicating technique. A special example with a two-piece power function and deterministic coefficients is presented to illustrate the general results. At last, explicit expressions of the optimal solutions are given.