This paper introduces the Trainable Feature Matching Attention Network (TFMAN) for Single Image Super-Resolution (SISR). TFMAN integrates Trainable Feature Matching (TFM) and Same-size-divided Region-level Non-Local (SRNL) modules to enhance the representation capabilities of Convolutional Neural Networks (CNNs) for SISR. TFM explicitly learns features from training images through feature matching, while SRNL reduces computational and memory demands by performing non-local operations in parallel on uniformly divided blocks of the input feature map. The proposed model uses a recurrent convolutional network as its backbone and is evaluated on benchmark datasets with degradation models (Bicubic, Blur-Downscale, and Downscale-Noise). TFMAN outperforms most state-of-the-art methods in terms of quantitative metrics (PSNR and SSIM) and qualitative visual quality, while using fewer parameters. Ablation studies and module explorations validate the effectiveness of TFM and SRNL. The code for TFMAN is available at \url{https://github.com/qizhou000/tfman}.This paper introduces the Trainable Feature Matching Attention Network (TFMAN) for Single Image Super-Resolution (SISR). TFMAN integrates Trainable Feature Matching (TFM) and Same-size-divided Region-level Non-Local (SRNL) modules to enhance the representation capabilities of Convolutional Neural Networks (CNNs) for SISR. TFM explicitly learns features from training images through feature matching, while SRNL reduces computational and memory demands by performing non-local operations in parallel on uniformly divided blocks of the input feature map. The proposed model uses a recurrent convolutional network as its backbone and is evaluated on benchmark datasets with degradation models (Bicubic, Blur-Downscale, and Downscale-Noise). TFMAN outperforms most state-of-the-art methods in terms of quantitative metrics (PSNR and SSIM) and qualitative visual quality, while using fewer parameters. Ablation studies and module explorations validate the effectiveness of TFM and SRNL. The code for TFMAN is available at \url{https://github.com/qizhou000/tfman}.