Pieter Pan's and PIGDOG's calculations are correct geometrically. A complicating factor is that neither light nor radio waves travel in perfectly straight lines in the atmosphere. The rays bend around the earth, increasing the distance to the visible or radio horizons by something like 10% more than the geometrical result.
Thus the rule of thumb tends to be
distance/nm = 1.2 * sqrt (height/ft)
(rather than the geometrical factor which is about 1.07)