The global health crisis caused by COVID-19 has highlighted the need for accurate diagnostic methods. Traditional testing methods face challenges due to their cost, the invasive nature of the tests, and potential errors. This research paper proposes densely convolutional neural network models for the classification of chest X-ray images, focusing on distinguishing between COVID-19, Normal, and Pneumonia cases. The study also compares the performance of different proposed Convolution Neural Network (CNN) architectures: DenseNet121, DenseNet169, and DenseNet201 with earlier reported work. Our analysis revealed that the proposed DenseNet169 outperformed the other models related to previous studies across key performance metrics. It achieved an impressive accuracy of 96% and the highest precision at 97%, while the proposed DenseNet121 and DenseNet201 also displayed high accuracy and low loss values, they did not match the overall balance of precision, recall, and F1-score provided by DenseNet169.