Laminar flame speeds and Markstein lengths of n-decane/air, Jet-A/air, and S-8/air flames were measured using spherically expanding premixed flames. The experiment used a spherical combustion chamber housed inside a customized oven, which provides a uniform temperature distribution inside the chamber for fuel evaporation. Linear and nonlinear extrapolation methods to obtain unstretched flame speed were compared. The difference between linear and nonlinear extrapolation is within 0.3-2.0 cm/s over the entire range of fuel equivalence ratio and decreases as the equivalence ratio increases. The measured flame speed data were compared to numerical simulations using several existing kinetic mechanisms and surrogate models. The results show that the JetSurF 0.2 mechanism was able to best represent the present measured flame speed data for n-decane. Surrogate models were simulated to represent the flame speeds of Jet-A and S-8. However, the simulated flame speeds overpredict the measured flame speeds. GC-MS and a negative ionization method were used to determine the composition change of the liquid fuels before and after heating/vaporization. The result shows that when the initial temperature is above 500 K, auto-oxidation occurred during the mixing process for Jet-A and S-8 flames, which likely produced ketones or aldehydes, resulting in falsely lower flame-speed data.