We simulated larval dispersal of the two most important commercial shrimp species (Penaeus indicus and Metapenaeus monoceros) on the Sofala Bank, Mozambique using a Lagrangian biophysical modelling tool Ichthyop (Lett et al. 2008), coupled to a Regional Ocean Modeling System (ROMS) configuration developed for the bank (Malauene et al. 2018). In total, 30 000 simulated larvae for the two species combined were released every 3 days for 5 model years from nine areas (including their true spawning areas (Malauene et al. 2021)), here referred to as five release areas comprising both inshore and offshore locations (Fig.1). For each simulation, individual larval trajectories were tracked for 15 days, the pelagic larval duration for these species. At the end of the simulation, shrimp larvae found outside the release areas (which are also the recruitment areas) were considered lost from the Sofala Bank population. To investigate the potential effect of cool water temperatures on larval mortality, we tested lethal temperature threshold values from 20 to 25oC for model year 5, i.e. one year for each of the six temperature simulations. In the simulations, larvae died when they encountered water cooler than the temperature thresholds. The present data are the processed simulation outputs of mean (with corresponding standard errors) percentages of larvae lost from all nine areas combined, the five release areas combined, and from each of the nine release areas, as well as larval mortality for each temperature simulation. The data include the total distances travelled along each trajectory and the linear distance from the origin to the final location for all larvae.