Digital watermarking has shown promise in protecting multimedia content effectively. Unfortunately, current watermarking techniques mainly target specific types of media, whereas content displayed on computer screens is often multi-modal and changes in real-time. Visual Screen Content (VSC) is particularly vulnerable to theft and leakage through covert and common computer screenshot operations, significantly limiting the protection scope and response time of mainstream watermarking techniques in screenshot scenarios. Furthermore, existing VSC watermarking methods struggle to balance preserving the visual quality of the original screen content with enhancing robustness against malicious attacks. To address these challenges, we propose a practical and robust watermarking method for arbitrary VSC protection, called ScreenMark. Specially, ScreenMark employs a three-stage progressive watermarking framework. First, inspired by diffusion principles, we initialize the mutual transformation between regular watermark information and irregular watermark patterns.Second, we utilize a pre-multiplication alpha blending rendering method to integrate watermark patterns with the screen, and pre-train a screen decoder to decode the corresponding watermark patterns. The progressively complex distorter enhances the robustness of the watermark on the surface in real-world screenshot scenarios and subsequent applications.Finally, we fine-tune the model under the guidance of the joint-level distorter. We establish a dataset containing 100,000 screenshots from various devices and resolutions. Extensive experiments on different datasets convincingly demonstrate the effectiveness in terms of robustness, invisibility, and practical applications.