In ‘Geographic profiling: The fast, frugal, and accurate way’ (Applied Cognitive Psychology, 2004, vol. 18, pp. 105–121), we demonstrated that most people are able to predict the home location of a serial offender by using a simple prediction strategy that exploits patterns found in the offender’s spatial behaviour. In this issue of Applied Cognitive Psychology, Rossmo challenges the validity of this research with respect to our data selection and methods of analysis. In response, we argue that: his proposed method for selecting data is unscientific; there is little evidence to support his claim that five crimes are required before profiles can be accurate; search area as a measure of profile accuracy has not yet been shown to be more useful than error distance; the heuristics we have examined are defined correctly and do lead to improvements in profile accuracy; and computerized geographic profiling is not a free service. Our comments aim to generate constraint in those intent on building confidence in computerized geographic profiling systems in the absence of strong empirical evidence to support their use.