A technique is presented whereby the standard error of a simple spatial average can be computed given any spatial configuration of measuring sites. The method should prove useful to those wishing to assess the time‐averaged performance of a network of rain gauges where the site locations are fixed within the averaging area. It is demonstrated that in addition to the site density and the spatial structure of the measured variable, the spatial geometry of the network with respect to the averaging area must be accounted for in the standard error expression. By incorporating estimates of the longterm population mean in the method, bias and random error in the sample statistics are minimized.