The present work explores the utility and value of geographical offender profiling methodologies within a novel context, considering both theoretical and practical issues relating to their application. The effectiveness of a well-known geographical profiling system, Dragnet, was tested across 101 New Zealand sex offence series, and findings compared with those derived for an equivalent sample from the UK. Average search costs (the amount of the total offence area that needed to be searched, starting from predicted offender home location, before the offender's actual home was reached) were far greater for the New Zealand sample than their UK offending counterparts. It is argued that this is because the spatial behaviour of New Zealand offenders violates many of the assumptions that Dragnet and other similar geographical profiling systems make in predicting offenders' home locations. Calibration of the system to the specific home-crime distance patterns of the New Zealand offenders did not enhance the efficacy of predictions made to a significant extent. It is consequently argued that, in their current form, geographical profiling systems are limited in their ability to account for samples displaying very different spatial characteristics to those that they were developed from and for. The implications of these findings for the general utility of geographical profiling are discussed, and ways in which systems might be developed in order to broaden their scope and applicability are suggested.