The current research experimentally studied flame speed and stretch under engine in-cylinder conditions. A direct-injection, spark-ignition, and optically accessible engine was utilized to image the flame propagation, and E10 was selected as the fuel. Also, three fuel–air mixtures (stoichiometric, lean, and rich) were examined. The flame front was located by processing high-speed images. This study introduces a novel approach for calculation of equivalent spherical flame radius for analysis of flame speed and stretch. Flame front propagation analysis showed that during the flame propagation period, the stretch decreased until the flame front touched the piston surface. This was a common trend for stoichiometric, lean, and rich mixtures, which occurred because the flame radius was the dominant factor in the stretch calculation. In addition, the rich fuel–air mixture showed a lower flame stretch compared to stoichiometric or lean mixture. This was the result of a lower Markstein number for the rich fuel–air mixture. To study the sensitivity of different fuel–air mixtures to the flame stretch, the trajectory of the flame centroid was tracked until the flame front touched the piston surface. The results showed that the end centroid for the lean mixture deviated from the start point more than those of the rich and stoichiometric mixtures because the lean mixture had a higher flame stretch and lower flame speed. Furthermore, comparing the flame stretch at three different engine speeds revealed that increasing the engine speed increased the flame stretch, especially during the early flame development period. According to previous studies which discussed flame stretch as a flame extinguishment mechanism, the probability of flame extinction is higher when the engine speed is higher. Also, uncertainty analysis was conducted to determine the effect of camera setting on the flame stretch. Results showed that a maximum relative uncertainty of 4.5% occurred during the early flame development.