The key challenge to integrate the unmanned aircraft system (UAS) into airspace is to develop a means to sense and avoid (SAA) other aircrafts. The main function of the SAA is self-separation, i.e. remaining “well clear” of the other aircrafts. The separation thresholds must be quantitatively defined for the UAS to autonomously maintain self-separation. In this paper, the separation thresholds are defined quantitatively for the UAS in a dynamic airspace full of aircrafts that differ in motion state and performance. Then, a “sector-like” dynamic collision-free region (CFR) was set up around the UAS. The size of the CRF can be adjusted adaptively according to the relative motion states of the surrounding intruders, the performance of the UAS, and the altitude of the airspace. The simulation results show that the proposed adaptive separation thresholds adapted to the dynamic airspace environment better than the fixed separation thresholds recommended by Sense & Avoid Science and Research Panel (SARP), and controlled the missing and false alarm rates on low levels. This means our adaptive separation thresholds can effectively balance the safety and operation efficiency of the airspace.