[jase (Jason)]

Indeed, I wouldn't be picking a very bright guide star. The brighter it is, the harder the algorithm needs to work to compute the centroid. As Leon points out, a star of medium brightness is best - certainly from an accuracy perspective.