Pulmonary disorders, such as pneumonia, tuberculosis, and COVID-19, have a significant global health impact, requiring reliable and swift diagnostic methods to improve treatment outcomes. This study examines the potential of a hybrid CNN-LSTM model to surpass conventional methods in accurately diagnosing various lung diseases from chest X-rays (CXRs). It introduces a novel CNN-LSTM model that utilizes spatial dependencies within 2D CXR images to enhance diagnostic accuracy. A collection of 4529 CXRs derived from Shenzhen, Montgomery, Belarus, and COVID-19 Radiography databases were used, encompassing a wide range of respiratory diseases. To improve image quality and minimize noise, Contrast Limited Adaptive Histogram Equalization (CLAHE) and median filtering were applied during pre-processing stage. The CNN component extracted hierarchical features, while the LSTM component treated image patches as sequential data to exploit spatial relationships and retain contextual information across the image. This hybrid approach enabled the model to learn complex representations for robust disease classification. Baseline CNN and CNN-BiLSTM models were also implemented for comparative evaluation. Results demonstrated that the CNN-LSTM model achieved the highest performance with an accuracy of 96.26% and precision, recall, and F1 scores of 96.44%, 96.62%, and 96.49%, respectively. It was followed by CNN-Bi LSTM and CNN models with accuracy rates of 95.58 and 94.24%. The outcome underscores the potential of the proposed CNN-LSTM model as a reliable tool for diagnosing pulmonary disorders from CXRs.