This paper describes an experimental activity performed on a passenger car powered by a spark ignition engine fuelled alternatively with natural gas (CNG) and hydrogen-natural gas blends, with 15% (HCNG15) and 30% (HCNG30) of hydrogen by volume. The vehicle was tested on a chassis dynamometer over different driving cycles, allowing the investigation of more realistic operating conditions than those examined on an engine test bed at steady state conditions. Fuel consumption was estimated using the carbon balance methodology, allowing the comparison of engine average efficiency over the driving cycles for the tested fuels. Furthermore, cylinder pressure was measured and, by processing the pressure signal; a combustion analysis was performed allowing to estimate the burning rate and combustion phasing. Ignition timing was the same for all the tested fuels, in order to assess their interchangeability on in-use vehicles. Results showed CO2 emission reduction between 3% and 6% for HCNG15 and between 13% and 16% for HCNG30 respect to natural gas. Fuel consumption in MJ/km did not show significant differences between CNG and HCNG15, while reductions between 3% and 7% have been observed with HCNG30. The heat release rate increased with hydrogen content in the blends, reaching values higher than those attained using CNG. The combustion duration, calculated as the angle between 10% and 90% of heat released, has been shortened, with 16% reduction for HCNG15 and 21% for HCNG30 respect to CNG at 2.5 bar imep and 2400 rpm. As a consequence, hydrogen addition resulted in a combustion phasing advance respect to CNG. Cycle-by-cycle variability decreased, particularly at low loads, due to the positive effect of hydrogen on combustion stability. Copyright (C) 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.