Understanding which factors most influence the level and type of pollution that is emitted from vehicle exhaust is important for environmental protection, health and transport planning. Most research in this area has been limited to a programmed drive cycle using a vehicle dynamometer to simulate real-world driving conditions. However, previous studies have shown that the exhaust emissions of carbon monoxide (CO) and hydrocarbons (HCs) are much lower on these standard simulated drive cycles than those measured under true real-world driving conditions or measured in the atmosphere at roadsides. In this study a vehicle was driven twice on a route in Sydney with one and two occupants. The output of the vehicle’s engine control unit and the composition of the vehicle’s exhaust were continuously logged while a voice recorder was used for positional information. Fuel consumption could be used to predict carbon dioxide (CO₂) emissions to a high degree of accuracy but not CO or HC emissions. This is because transient high load operation for this vehicle can induce the occurrence of “enrichment events” where the air to fuel ratio is richer than stoichiometric and therefore results in dramatic increases in these emissions. A good predictor of these events was a throttle opened 29° or more, for which CO and HC emissions increase by at least an order of magnitude. Emission factors, both fuel and distance based, are reported for the entire route as well as selected road links and categorised by pre- or post-catalytic converter warm-up).