The aim of this work is to provide insight into the state-of-the-art turbulent flame speed (ST) correlations and to determine the most appropriate correlations to use under different turbulent premixed combustion conditions. The accuracies of 16 correlations for ST are determined using a large volume of data over a range of conditions. Accuracy is based on a mean absolute percentage error (MAPE). The comparison is completed once using the original authors' adjustable parameters and a second time using parameters proposed by the current work that minimize MAPE for four different groups of data. Based on the results of the analysis using the newly-suggested parameters, the five most accurate correlations are then further examined to evaluate their respective abilities to predict trends under various turbulent conditions. While many correlations perform well over the range of data (MAPE < 33%), no single correlation can predict all experimentally-observed trends for methane flames under these conditions. Further issues are found when predicting trends for larger hydrocarbons; ethane and propane. Although low errors are again found (MAPE < 25%), correlations are not generally able to replicate the observed trends of experimental data for C2H6 and C3H8. While it is commonly accepted that no single correlation can accurately predict Sr, this work has shown that the correlation derived by Muppala provides the closest overall agreement to the data examined. However it cannot be defined as a general correlation. For this reason the authors have proposed to continue development of an ST modeling tool based on a modified version of the Cantera 1D freely propagating laminar flame speed (S-L) code. Greater cooperation in the ST research community to expand the range of available experimental data and better enable direct comparison of data and correlations from different experiments is also recommended.