How do you find the best measure of central tendency when using outliers?

A weighted mean is probably best. Certainly better than a median which throws away information from most of the observations.