Multimodal Cross-Lingual Summarization (MXLS) which aims to generate target-language (e.g., Chinese) summaries with multimodal inputs, e.g., source-language (e.g., English) document and visual features, has attracted much attention in the research community. Existing work mainly focuses on distilling knowledge from multimodal monolingual summarization (MMS) to MXLS due to the close relationship between them or multi-task learning to enhance the performance of MXLS. However, these methods rely on large-scale data and do not fully exploit the potential of MMS, which is especially not beneficial for low-resource languages. Besides, the visual features may be not summary-related, and thus directly using them hurts the model’s performance. To address the issue, we propose a novel sequential learning framework (SLF) for MXLS. Specifically, we utilize one unified decoder to generate the sequential concatenation of monolingual and several cross-lingual summaries, and thus our method makes the MMS task a prerequisite of the MXLS task and fully exploits the knowledge across languages. In this manner, the shared decoder learns interactions involving alignments and summary patterns across languages, which encourages explicit knowledge transfer from MMS to MXLS. Furthermore, we propose a dynamic way to adaptively incorporate the visual features. Extensive experiments on many-to-many settings show the effectiveness of the proposed SLF model.