This paper describes an attempt made to study the time-headway distribution of urban heterogeneous traffic over a wide range of flow levels. An appropriate methodology for the collection and extraction of headway data has been suggested. The importance of a systematic procedure for grouping of headway data through appropriate choice of class interval has been discussed. The methodology for choice of best fitting statistical distribution to the observed headway data has also been described. The result of the study has shown that, the headways of urban mixed-traffic dominated by smaller vehicles like motorised two-wheelers can be modelled using negative exponential distribution over a wide range of traffic flow levels.