This paper proposes OSEDiff, a one-step effective diffusion network for real-world image super-resolution (Real-ISR). The key idea is to directly use the low-quality (LQ) image as the starting point for diffusion, eliminating the need for random noise sampling. This approach reduces computational cost and improves the stability of the output. The pre-trained diffusion model is fine-tuned with trainable LoRA layers to adapt to complex image degradations. Variational score distillation (VSD) is applied in the latent space to ensure that the generated images align with natural image priors. The proposed method achieves comparable or better performance than existing multi-step diffusion-based methods in both objective metrics and subjective evaluations. The model is efficient, with significantly reduced inference time and fewer trainable parameters. The experiments demonstrate that OSEDiff can generate high-quality images in just one diffusion step, making it a promising solution for Real-ISR tasks. The source code is available at https://github.com/cswry/OSEDiff.This paper proposes OSEDiff, a one-step effective diffusion network for real-world image super-resolution (Real-ISR). The key idea is to directly use the low-quality (LQ) image as the starting point for diffusion, eliminating the need for random noise sampling. This approach reduces computational cost and improves the stability of the output. The pre-trained diffusion model is fine-tuned with trainable LoRA layers to adapt to complex image degradations. Variational score distillation (VSD) is applied in the latent space to ensure that the generated images align with natural image priors. The proposed method achieves comparable or better performance than existing multi-step diffusion-based methods in both objective metrics and subjective evaluations. The model is efficient, with significantly reduced inference time and fewer trainable parameters. The experiments demonstrate that OSEDiff can generate high-quality images in just one diffusion step, making it a promising solution for Real-ISR tasks. The source code is available at https://github.com/cswry/OSEDiff.