AbstractThe ground‐water observation well network in many parts of Kansas has been developed and expanded through the years without serious attempt to determine the adequacy of the network for any specified purpose or to assess its cost effectiveness. This study was undertaken to examine the existing well network in northwest Kansas and to determine the arrangement that offers the most satisfactory accuracy for the purpose of monitoring it. To achieve this goal, we have employed the theory of regionalized variables to estimate the amount of spatial variability of the water table. The error analysis produced by universal kriging indicates that a significant reduction in the number of wells could be achieved by employing a regular 4‐mile (6.4‐km) network, without affecting the present level of accuracy. It also indicates that it is not practical to reduce the estimation error in the water‐table surface uniformly throughout the region because to do so would increase the cost of monitoring wells drastically. For example, reducing the presently existing error by 50 percent throughout the area would require 16 times more wells than the currently existing well