A systematic computational study is conducted to understand the propagation of uncertain chemical kinetic parameters into turbulent premixed flame simulations. Two snapshots of the cross-section of a spherical flame kernel in homogeneous isotropic turbulence is extracted from a three dimensional direct numerical simulation (DNS) and serve as the initial conditions for a series of subsequent two-dimensional DNS that sample and quantify the impact of chemical kinetic uncertainties on the integrated heat release rates and certain species concentrations. The target configuration features real jet fuel chemistry modeled by a 31-species reduced reaction model. The importance of each reaction is first ranked, by comparing the normalized reaction flux based on a snapshot from the DNS. A two-step process is then taken where initial Monte Carlo screening is performed by perturbing 99 reaction rate constants simultaneously. Following this, a subset of 19 reactions is studied using a larger collection of 2D DNS simulations. As the major quantity of interest, the integrated heat release rate is consistently the most sensitive to reactions R8 (H + O-2 = O + OH) and R32 (CO + OH = CO2 + H). When combined with a consideration of the reaction rate uncertainty, R71 (CH3 + OH = CH2* + H2O) consistently impacts the quality of flame prediction most prominently. For the case starting with a strong burning flame, the rate limiting reactions are found to be not significantly different between the turbulent and laminar conditions, although the magnitudes of sensitivity coefficients are amplified for certain reactions under turbulent conditions due to an enlarged thermochemical space. For the case initialized with a near-extinction flame, more upstream and/or competing reactions with R8 and R32 are observed. Another manifestation of the enlarged thermochemical space in turbulent flames is the wider range of influential reactions for integrated mass of CO, in comparison with the laminar condition where the sensitivity is dominated by R32. For H and C2H2, reactions R71, R42 (HCO + O-2 = CO + HO2) and R40 (HCO + M = CO + H + M) are identified to impact the model predictions for such species as acetylene in the most significant manner. (C) 2019 The Combustion Institute. Published by Elsevier Inc. All rights reserved.