Direct-injection spark-ignition engines have become popular due to their flexibility in injection strategies and higher efficiency; however, the high-pressure in-cylinder injection process can alter the airflow field by momentum exchange, with different effects for fuels of diverse properties. The current paper presents results from optical studies of stoichiometric combustion of ethanol, butanol, iso-octane and gasoline in a direct-injection spark-ignition engine run at 1500 RPM with 0.5 bar intake plenum pressure and early intake stroke fuel injection for homogeneous mixture preparation. The analysis initially involved particle image velocimetry measurements of the flow field at ignition timing with and without fuelling for comparison. Flame chemiluminescence imaging was used to characterise the global flame behaviour and double-pulsed Laser-sheet flame tomography by Mie scattering to quantify the local topology of the flame front. The flow measurements with fuel injection showed integral length scales of the same order to those of air only on the tumble plane, but larger regions with scales up to 9 mm on the horizontal plane. Averaged length scales over both measurement planes were between 4 and 6 mm, with ethanol exhibiting the largest and butanol the smallest. In non-dimensional form, the integral length scales were up to 20% of the clearance height and 5-12% of the cylinder bore. Flame tomography showed that at radii between 8 and 12 mm, ethanol was burning the fastest, followed by butanol, iso-octane and gasoline. The associated turbulent burning velocities were 4.6-6.5 times greater than the laminar burning velocities and about 13-20% lower than those obtained by flame chemiluminescence imaging. Flame roundness was 10-15% on the tomography plane, with largest values for ethanol, followed by butanol, gasoline and iso-octane; chemiluminescence imaging showed larger roundness (18-25%), albeit with the same order amongst fuels. The standard deviation of the displacement of the instantaneous flame contour from one filtered by its equivalent radius was obtained as a measure of flame brush thickness and correlated strongly with the equivalent flame radius; when normalised by the radius, it was 4-6% for all fuels. The number of crossing points between instantaneous and filtered flame contour showed a strong negative correlation with flame radius, independent of fuel type. The crossing point frequency was 0.5-1.6 mm(-1). The flame brush thickness was about 1/10th of the integral length scale. A positive correlation was found between integral length scale and flame brush thickness and a negative correlation with crossing frequency. (C) 2015 The Combustion Institute. Published by Elsevier Inc. All rights reserved.