As stated in the Read Me file: These data and resources are associated with the manuscript: Learning to count: determining the stoichiometry of bio-molecular complexes using fluorescence microscopy and statistical modelling, Mersmann et. al., as submitted to biorXiv in July 2020. The raw imaging data relates to Figure 5, S1, S2 and Table S1. The images are fluorescent micrographs displaying immobilised adenovirus particles bound to a monoclonal antibody 9C12. Each experiment folder is numbered, as in Table S1, and appended with the mixing proportion (Fl), as defined in the manuscript. Within each folder there are 6 subfolders, representing samples incubated with different concentrations of 9C12 antibody. Each image is a 3 channel 1024x1024 tif. Channel 1 = 9C12 Alexa Fluor 647. Channel 2 = 9C12 Biotin + QDot655. Channel 3 = Adenovirus Alexa Fluor 488. Samples were illuminated in TIRF mode using a 100X objective, images were captured on a Hamamatsu OCRA Flash 4 sCMOS camera. Further details are available in the header of each file. The control samples are labelled with 100% 9C12 Alexa Fluor 647 or 100% 9C12 Biotin, as described in the manuscript. The data analysis script is an imageJ macro. It runs on the FIJI version of ImageJ with the NanoJ package installed (https://github.com/HenriquesLab). It outputs fluorescent measurements for each identified AdV particle. Note that the script rearranges the channel order such that Channel 1 = Adenovirus Alexa Fluor 488, Channel 2 = 9C12 Alexa Fluor 647, Channel 3 = 9C12 Biotin + QDot655. The channels require registration due to chromatic aberration, this is achieved using the Realign Channels function in NanoJ, appropriate translation masks are provided along with the script. Any question about the data or script should be addressed in Joe Grove (j.grove@ucl.ac.uk)