The invention discloses a long-term deflection monitoring data based concrete bridge stiffness reliability evaluation method. The method comprises the following steps: 1) preprocessing monitored bridge deflection data to obtain a real bridge deflection time interval x; 2) performing peak identification on the deflection time interval x obtained in the step 1 to obtain a peak sequence p; 3) according to the peak sequence p, performing probability distribution fitting of a deflection peak, and performing fitting check to obtain a probability density function, a mean value and a variable coefficient of the deflection peak; 4) establishing a function equation Z for calculating the bridge stiffness reliability; and 5) calculating a reliability index beta. According to the method, the function equation Z not only comprises actual operation information of a bridge but also covers design parameters of the bridge; and the method is used for evaluating the reliability index of an actual stiffness level of the bridge relative to an ideal design state, and is especially suitable for evaluating the bridge state based on the deflection data in bridge health monitoring.