Two fundamental aspects of LGS references in Adaptive Optics, namely the conical anisoplanatism and the tilt indetermination, have been attacked in the last years by several authors and using several approaches. It turns out that tilt recovery schemes are plagued by some apparently unavoidable conical anisokinetism error. We have shown in this paper that, one time the problem is faced in a global manner a simple solution to the conical anisokinetism error can be obtained.