Rapid compression machines (RCMs) have often been used to study auto-ignition phenomena and to validate chemical kinetic models at conditions relevant to engines. Many RCM designs were historically used for the latter purpose. First, the flat piston configuration was extensively used and contributed to valuable ignition delay data. Temperature heterogeneity in these devices was later addressed, and new designs were proposed. In particular, creviced piston configurations have been demonstrated to enhance the temperature homogeneity in the bulk volume of RCM chambers. Nevertheless, data from flat piston RCMs are still used in the community. Thus, additional understanding of experimental results from such devices is still useful, and it also helps to enhance the understanding of temperature stratification effects on auto ignition phenomena. This paper addresses the interpretation of measured ignition delays from a flat piston RCM (without a circumferential crevice in the piston). First, we characterize local and temporal temperature evolution with a toluene planar laser induced fluorescence (PLIF) technique applied in oxygen-free gas under different test conditions. At a compression pressure (P-c) of 11 bar and an adiabatic core temperature (T-c) of 750 K, we find that the maximum measured temperature is very close to T-c up to 30 ms after the compression stroke. In the second step, we qualitatively compare temperature fields with chemiluminescence images during the auto-ignition of n-hexane and an n-heptane/methyl-cyclohexane (MCH) blend for a fuel-air equivalence ratio of 0.4, P-c values of 11 bar and 16 bar and T-c values of 700-900 K. The chemiluminescence results of n-hexane show a fast sequential auto-ignition which strongly suggest that the combustion propagation regime is mainly a reaction fronts regime. In the n-heptane/MCH case, the tracking of ignition kernels demonstrates that the auto-ignition initiates in the cold vortex at 71 values of 787 K and 834 K and in the adiabatically compressed region at T-c values of 742 K and 900 K. In this study, flat piston RCM experiments demonstrate temperature heterogeneities with ranges up to 120 K and gradients of approximately 160 K/mm. This significantly influences the ignition delay measurement in the negative temperature coefficient (NTC) range. Through the example of n-heptane/MCH auto-ignition, it seems that despite temperature heterogeneities, the lower limit of the NTC range is well detected, whereas the higher limit is over-estimated. The determination of temperature thresholds in which ignition delay data may be confounded highly depends on the stratification level and on complex thermo-kinetic interactions. Outside these thresholds, when combustion is dominated by auto-ignition in regions at the adiabatic core temperature, the experimental data are easier to compare with simulation frameworks assuming a homogeneous hypothesis. In chemical kinetics-oriented studies, temperature heterogeneity should be considered when flat piston RCMs are used and when the piston crevice is not adequate. (C) 2016 Elsevier Ltd. All rights reserved.