Abstract Previous measurements of turbulent burning velocity ( S T ) have been reported by Gulder and colleagues for intense levels of turbulence, defined to be u ’⁄ S L values between 12 and 24, and normalized integral scales ( L x ⁄ δ L ) up to 46. The present work extends burning velocity measurements to much higher levels of turbulence than have been considered before: to extreme turbulence defined as u ’⁄ S L values from 25 to 163 and L x ⁄ δ L up to 114. These conditions are argued to be more representative of the turbulence found in certain engines. To do so, a new large, piloted Bunsen burner (called Hi-Pilot) was developed and OH and formaldehyde PLIF images provided the time-averaged contours of progress variable based on OH ( c OH ). The conventional global consumption speed ( S T , GC ,1 ⁄ S L ) is based on the c OH = 0.5 contour and it was found to exceed 25. Two other measured speeds are based on the leading edge ( S T , GC ,2 ) and the component due to flamelet surface density ( S T , F ). Varying the integral scale had a significant effect on S T , GC ,2 but not on the other two burning velocities. The consumption speed S T , GC ,1 curve displayed “bending” in the range of extreme turbulence, while the flamelet surface density contribution ( S T , F ) curve instead flattened out and was independent of turbulence intensity. A possible explanation for these measured trends is based on the observed extensive broadening of the preheat zone. Preheat broadening depends on the integral scale and is believed to attenuate the turbulence that eventually interacts with the reaction zone. Results indicate a breakdown of the laminar flamelet assumption; it appears that preheat broadening may cause thermal diffusivity to dominate over the flame wrinkling mechanism.