CaFFlow is a Python framework designed for the acquisition and analysis of a variety of image/video streams, including calcium imaging data recorded by small head mounted microscopes (e.g. Miniscope), two-photon microscopes, and/or experimental subject’s behavior recorded by generic video sources. Compared to existing software, the main advantage of the CaFFlow framework is its modular structure that allows creation of ‘processing pipelines’ tailored to particular research projects while keeping strict standards on the usage of computational algorithms. This approach ensures the transparency and reportability of the analysis and prioritizes reproducibility over approaches requiring tuning of a large parameter space. The general concept behind the CaFFlow framework is the well-known idea of representing a video stream as a ‘flow’ of images or video frames generated by a ‘frame source’, propagated over a directed graph of ‘frame processors’ and being stored into a ‘frame sinks’. This architecture supports a wide-range of tasks, ranging from simple video format/size conversion and subject's position detection to dF/F calcium trace extraction and pre-processing of data generated by high throughput automated whole organ imaging system such as TissueCyte.