In this paper we introduce an online handwritten document database (CASIA-onDo), serving as a standard database for the devel opment and evaluation of methods in the field of online handwritten document layout analysis. It consists of 2,012 documents including a total of 841,159 online strokes. The database, covering Chinese and En glish languages, was produced by 200 writers. Six types of contents occur in the documents, namely text, formulas, diagrams, tables, figures, and lists. The distribution of different types is close to the actual situation. Benefiting from detailed annotations, CASIA-onDo can support differ ent tasks of layout analysis under online or offline settings. Firstly, based on the semantic level annotation, it can be used for many classification tasks such as text/non-text classification, table/non-table classification, multi-class stroke classification and so on. Secondly, based on the in stance level annotation, it can be used for segmentation tasks such as text line separation and formula segmentation. Thirdly, based on the various writing styles, it can be used for handwriting recognition and writer clustering tasks. In addition, we perform preliminary experiments to provide a benchmark on this database with a state-of-the-art method. More techniques can be evaluated on this challenging database in the future.