The primary problem addressed in this study is the challenge of accurately classifying hand motions using Electromyography (EMG) signals, which are inherently complex due to the overlapping electrical activities of flexor and extensor muscles. The aim of the research is to develop a robust classification model for ten distinct hand gestures by employing autoregressive feature extraction in combination with various classifiers, including K-Nearest Neighbors (KNN), Support Vector Machine (SVM), and Linear Discriminant Analysis (LDA). The methodology involved recording EMG signals from two channels and applying autoregressive modeling to extract relevant features. The classifiers were then evaluated for their performance based on classification accuracy. The results revealed that LDA achieved the highest accuracy of 80% at an autoregressive order of 12, while KNN’s performance declined to an accuracy of 50%, indicating its sensitivity to feature complexity. Additionally, the confusion matrix showed that while certain gestures, such as “CLO,” were classified perfectly, others like “IND” and “THI” experienced misclassification rates of 30% and 46%, respectively. In conclusion, while LDA proved effective in classifying EMG signals for hand gestures, the variability in accuracy highlights the need for further refinement in feature extraction and potentially the exploration of more advanced classifiers to improve recognition capabilities in practical applications such as prosthetics and rehabilitation engineering.