This dataset contains a series of three-dimensional fluorescence microscopy videos of the rat cardiomyoblast cell-line H9c2. The outer mitochondrial membrane is labeled with either mCherry (a red fluorescent protein) or eGFP (enhanced green fluorescent protein). Membrane-bound vesicles called endosomes are labeled using mCLING-ATTO647n. A portion of the data is further labeled for acidic vesicles (LysoTracker) and/or has correlative images in brightfield mode (label-free, transmission microscopy). This data is a collection of fluorescence microscopy videos acquired to study the dynamics of mitochondria and the interaction of different subcellular vesicles. Correlative transmission microscopy images (trans) were provided for machine learning to later enable label-free detection of mitochondria and distinguish between different types of vesicles. As three different cameras were used to enable the fast multi-color acquisition, all data has been "channel aligned", i.e., registered to the same sample coordinate system for accurate colocalization of different cellular components. Both unprocessed widefield images and deconvolved data are provided (dimensions: 3D-channels-time) plus maximum intensity z-projected data and AVI-movies of the deconvolved images to enable a quick and easy visualization of the else large image files.