MedMamba is a novel architecture designed for generalized medical image classification, combining convolutional layers and state space models (SSMs) to effectively model long-range dependencies while maintaining linear computational complexity. The proposed SS-Conv-SSM block integrates convolutional layers for local feature extraction with SSMs for capturing long-range dependencies, enabling efficient and accurate classification of medical images from various modalities. Extensive experiments on 16 datasets with ten imaging modalities and 411,007 images demonstrate that MedMamba outperforms state-of-the-art methods in terms of accuracy and efficiency. The model's performance is validated through various metrics, including overall accuracy, precision, sensitivity, specificity, F1-score, and AUC. MedMamba also shows robustness to image perturbations and provides valuable insights into its decision-making process through visual interpretation techniques like Grad-CAM and t-SNE. The source code and pre-trained weights are available at <https://github.com/YubiaoYue/MedMamba>.MedMamba is a novel architecture designed for generalized medical image classification, combining convolutional layers and state space models (SSMs) to effectively model long-range dependencies while maintaining linear computational complexity. The proposed SS-Conv-SSM block integrates convolutional layers for local feature extraction with SSMs for capturing long-range dependencies, enabling efficient and accurate classification of medical images from various modalities. Extensive experiments on 16 datasets with ten imaging modalities and 411,007 images demonstrate that MedMamba outperforms state-of-the-art methods in terms of accuracy and efficiency. The model's performance is validated through various metrics, including overall accuracy, precision, sensitivity, specificity, F1-score, and AUC. MedMamba also shows robustness to image perturbations and provides valuable insights into its decision-making process through visual interpretation techniques like Grad-CAM and t-SNE. The source code and pre-trained weights are available at <https://github.com/YubiaoYue/MedMamba>.