The p-median model locates facilities to provide optimal service to target populations. Where, for some reason, an inappropriate variable is used to stand in for a target population's demand, less than ideal facility systems can result. This effect is termed surrogation error. In this paper, I introduce this concept and perform an experiment which, with data for 25 Canadian cities, demonstrates that significant surrogation error can occur if general population is used in place of children or senior populations. I identify some of the correlates of surrogation error and conclude with a warning to location scientists to be conscious of, and to try to avoid, this problem.