Abstract: This article focuses on variability of the couplings between surface air temperatures (SAT) and northern annular modes (NAM) at various levels. To measure the coupling intensity of whole SAT and NAM anomaly fields, the coupling index has been defined as the leading principal component of the Partial Least Squares Regression model of SAT anomaly, and both the composite analysis and the coupling index have been used to reveal level-by-level and monthly variability of the coupling relations between upper anomalous northern annular mode and surface air temperature in the northern hemisphere. The major results are as follows: January SAT anomalies are stronger coupled with the January NAM anomaly at the middle-upper tropospheric levels than those at the other levels, while February SAT anomalies with the January NAM anomaly at the lower stratospheric level. The January NAM anomaly at the middle-upper tropospheric levels is strongest coupled with January SAT anomaly, and the coupling intensity is successively reduced month by month and becomes trivial after April; and the January NAM at the lower stratospheric level is stronger coupled with January as well as February and March NAM anomalies, but the coupling becomes trivial after April.