Understanding and monitoring the changes in the cognitive workload of trainees can offer critical quantitative information about their progression and performance. Unfortunately, accurate real-time objective quantification of cognitive workload has, thus far, proven elusive and is often neglected in favor of subjective self-reports. This paper reports a novel technique for the classification of cognitive workload using methods from the domain of deterministically nonlinear dynamical systems. The reported technique utilizes physiological input data, specifically the subject's electrocardiographic (ECG) signal, captured during task performance. The novelty of the proposed algorithm stems from its ability to perform real-time, as well as after-action review, classification of cognitive workload using the full ECG signal. As will be presented, the use of the full ECG signal offers the ability to determine even small changes in the subject's workload and proves itself far more accurate than the standard classification methodology using heart rate variability (HRV). Further, the proposed methodology offers the ability to create accurate, real-time workload metrics over diverse populations and tasks; thus, reducing the need for individualized model creation. The proposed algorithm is validated through a case study in which participants were asked to perform varying levels of the Multi-Attribute Task Battery (MATB) developed by NASA. The case study punctuates the high accuracy of the proposed algorithm and its ability to classify cognitive workload levels in real-time and after-action review.