Purpose: This study aimed to determine the accuracy of a 4 split time modelling method to generate velocity-time and velocity-distance variables in elite male 100-m sprinters and subsequently to assess the roles of key sprint parameters with respect to 100-m sprint performance. Additionally, this study aimed to assess the differences between faster and slower sprinters in key sprint variables that have not been assessed in previous work. Methods: Velocity-time and velocity-distance curves were generated using a mono-exponential function from 4 split times for 82 male sprinters during major athletics competitions. Key race variables-maximum velocity, the acceleration time constant (tau), and percentage of velocity lost (nu(Loss))-were derived for each athlete. Athletes were divided into tertiles, based on 100-m time, with the first and third tertiles considered to be the faster and slower groups, respectively, to facilitate further analysis. Results: Modelled split times and velocities displayed excellent accuracy and close agreement with raw measures (range of mean bias was -0.2% to 0.2%, and range of intraclass correlation coefficients (ICCs) was 0.935 to 0.999) except for 10-m time (mean bias was 1.6% +/- 1.3%, and the ICC was 0.600). The 100-m sprint performance time and all 20-m split times had a significant near-perfect negative correlation with maximum velocity (r >= -0.90) except for the 0 to 20-m split time, where a significantly large negative correlation was found (r = -0.57). The faster group had a significantly higher maximum velocity and tau (p < 0.001), and no significant difference was found for nu(Loss) (p = 0.085). Conclusion: Coaches and researchers are encouraged to utilize the 4 split time method proposed in the current study to assess several key race variables that describe a sprinter's performance capacities, which can be subsequently used to further inform training.