The purpose of the study was to test the potential utility of a geographical profiling approach for three separate series of rapes committed by a single offender. Two different mathematical distance-decay functions using either a normal distribution with a mean distance and standard deviation based on previous research or a truncated negative exponential function based on distances between crime sites in the series under investigation were applied to each of the series giving prioritised search areas the accuracy of which was then assessed. The prioritised area that had to be searched before the home base of the offender could be located varied from 7.60 km(2) or 2.15% of the total search area at its best to 15 1.10 km(2) or 42.66% at its worst for the normal distribution based on previous research and from 42.06 km(2) or 11.88% to no improvement when the truncated negative exponential function was used. The functions used showed less predictive ability when the offender was a commuter. Explanations for the variations in the findings as well as suggestions for improvements were outlined in the discussion.