Various phenomenological studies indicate that the mixing angle $��_{K_1}$ of $K_{1A}$ and $K_{1B}$, the strange partners of the axial-vector mesons $a_1(1260)$ and $b_1(1235)$, respectively, lies in the vicinity of $35^\circ$ or $55^\circ$, but whether this angle is larger or smaller than $45^\circ$ still remains controversial. When the $f_1(1285)$-$f_1(1420)$ mixing angle $��_{^3P_1}$ and the $h_1(1170)$-$h_1(1380)$ mixing angle $��_{^1P_1}$ are determined from the mass relations, they depend on the masses of $K_{1A}$ and $K_{1B}$, which in turn depend on the mixing angle $��_{K_1}$. We show that the approximate decoupling of the light $q\bar q$ state from the heavier $s \bar s$ state, which is empirically valid for vector, tensor and $3^{--}$ mesons, when applied to isoscalar axial-vector mesons, will enable us to discriminate different solutions of $��_{^3P_1}$ and $��_{^1P_1}$ and pick up $��_{K_1}\sim 35^\circ$. Indeed, for $��_{K_1}\sim 55^\circ$, the predicted $��_{^1P_1}$ disagrees sharply with the recent lattice calculation and the implied large $s\bar s$ content of $h_1(1170)$ and $q\bar q$ component of $h_1(1380)$ cannot explain the observation of their strong decays. We conclude that $��_{K_1}$ is smaller than $45^\circ$.

9 pages