In order to detect turn-to-turn fault sensitively, a protection scheme for turn-to-turn fault based on the particle swarm optimization (PSO) fitness is proposed according to the influence of turn-to-turn fault on equivalent circuit of transformers. Firstly, the fitness function for resistance parameter identification is constructed on the basis of the active loss equation of a transformer. Then, the parameter identification is achieved by PSO, and the minimum fitness value of search space is obtained. Finally, based on the minimum fitness value, a criterion is constructed to identify turn-to-turn fault. Taking transition resistance and measurement noise into account, the effectiveness of this protection is verified in simulations. Moreover, the factors that possibly affect the sensitivity of this protection are analyzed. In addition, the sensitivity comparison of this proposed protection and the protection principle based on equivalent circuit equilibrium equation and excitation inductance is carried out. Some dynamic simulation experiments are accomplished at last. The conclusions can be drawn that this proposed protection scheme can achieve high sensitivity in turn-to-turn fault detection and avoid the impact of parameter calculation error; the identification effect is not affected by load condition, and the detection sensitivity is still high under the influence of a certain amount of transition resistance; the detection effect is better than the other two protections mentioned above. The dynamic simulation results verify the effectiveness of the proposed protection scheme.