As an aid to survey design, we used data acquired from three European continuous GPS networks to test the precision of position estimates from static observations as a function of the length of the observing session and the number and distribution of reference stations. Our criterion was the weighted RMS of estimates over 31 days with respect to coordinates determined from 24-h sessions over a 2-year period. With a single reference station, a precision of 3 mm horizontal and 10 mm vertical could be achieved reliably only for session lengths of 3 h or longer and baselines less than 200 km. If four or more reference stations are used, these levels of precision were usually achieved with sessions as short as 2 h. With sessions 6 h or longer and four or more reference stations, the precision is typically 1---2 mm in horizontal and about 3---5 mm in vertical. Increasing the number of reference stations further provides only marginal improvement. Although there is some variation in precision in 4-station networks with the choice of reference stations, the dependence on distance and geometric distribution is weak.