This dataset contains time-lapse microscopy recordings (.ome.tif) of single-molecule translation events in HeLa cells, acquired for the study “Single-mRNA imaging and modeling reveal coupled translation initiation and elongation rates” (Lamberti, Chao, Gobet, Naef, 2025). The recordings were obtained using the SunTag system to fluorescently tag nascent peptide chains, allowing direct observation of ribosome dynamics on single mRNA molecules. The dataset includes recordings from four different SunTag reporters, each stably integrated into a common HeLa background with actin-anchored MS2-tagged mRNAs and scFv-GFP for nascent peptide labeling. The reporters differ by their coding sequence insert in the second half of the ORF: No-insert: SunTag backbone only without any insert. Renilla: Renilla Luciferase. PPG: insert with proline-proline-glycine (COL1A1-derived) motifs. AAG: same as PPG but with prolines mutated to alanines. Each reporter was imaged in two conditions: Steady-state translation: to assess bursting dynamics and ribosome density. Harringtonine (HT) run-off: to estimate elongation and initiation rates by observing ribosome clearance post-HT treatment. Images were acquired on a spinning-disk confocal microscope at 20-second intervals for up to 90 minutes, at the basal cell membrane to ensure high mRNA visibility and spatial anchoring. Fluorescence data were processed using flat-field and dark-current corrections, and intensity traces were quantified via Bayesian inference. These time-lapse traces formed the basis of a Totally Asymmetric Exclusion Process (TASEP)-based Hidden Markov Model that decodes ribosome number and kinetic rates from the intensity dynamics of single mRNAs (Github).