<p>Gait-based emotion recognition (GBER) uses walking patterns to detect emotional states, developments in sensing technologies and deep learning (DL) have expanded research, requiring a systematic review of existing methods, datasets, and performance trends. The study analyses GBER techniques, feature extraction, classification models, datasets, and real-world applications to highlight advancements, challenges, and future directions. This systematic literature review (SLR) follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses(PRISMA) guidelines to ensure transparency and reproducibility. Relevant studies were collected from reputed journals and conferences using a structured review protocol. An initial set of 411 articles underwent a multi-stage screening process that included duplicate removal, title and abstract screening, full-text evaluation, application of predefined search strings, and quality assessment. Based on these criteria, 50 studies were selected for in-depth analysis to examine methodological trends, reported recognition accuracies, dataset usage patterns, research growth, and validation practice. Deep Learning (DL) methods, appeared in (47%) of the studies, followed by hybrid models contributing to (31%) of the articles; machine learning approaches were used in (22%) of the studies. Considering the datasets used in this domain, researchers focused on using the Emotion-Gait (24.10%), followed by Carnegie Mellon University- Motion Capture Database (CMU-MoCap) and Edinburgh Locomotion Mocap Dataset (ELMD) with DFD (12.20%).The studies show that the highest average accuracies (89.50%) were achieved by graph convolution neural networks(GCNS), followed by Spatio temporal graph convolution neural networks(ST-GCN) achieving 88% accuracy.</p>