Abstract This paper presents a disaggregate approach to crash rate analysis. Enumerating crash rates on a per trip-kilometer basis, the proposed method removes the linearity assumption inherent in the conventional quotient indicator of accidents per unit travel distance. The approach involves combining two disparate datasets on a geographic information systems (GIS) platform by matching accident records to a defined travel corridor. As an illustration of the methodology, travel information from the Victorian Activity and Travel Survey (VATS) and accident records contained in CrashStat were used to estimate the crash rates of Melbourne residents in different age–sex groups according to time of the day and day of the week. The results show a polynomial function of a cubic order when crash rates are plotted against age group, which contrasts distinctly with the U-shape curve generated by using the conventional aggregate quotient approach. Owing to the validity of the many assumptions adopted in the computation, this study does not claim that the results obtained are conclusive. The methodology, however, is seen as providing a framework upon which future crash risk measures could be based as the use of spatial tracking devises become prevalent in travel surveys.