SUMMARYTo understand the mechanics of fish swimming, we need to know the forces exerted by the fluid and how these forces affect the motion of the fish. To this end, we developed a 3-D computational approach that integrates hydrodynamics and body dynamics. This study quantifies the flow around a swimming zebrafish (Danio rerio) larva. We used morphological and kinematics data from actual fish larvae aged 3 and 5days post fertilization as input for a computational model that predicted free-swimming dynamics from prescribed changes in body shape. We simulated cyclic swimming and a spontaneous C-start. A rigorous comparison with 2-D particle image velocimetry and kinematics data revealed that the computational model accurately predicted the motion of the fishʼs centre of mass as well as the spatial and temporal characteristics of the flow. The distribution of pressure and shear forces along the body showed that thrust is mainly produced in the posterior half of the body. We also explored the effect of the body wave amplitude on swimming performance by considering wave amplitudes that were up to 40% larger or smaller than the experimentally observed value. Increasing the body wave amplitude increased forward swimming speed from 7 to 21 body lengths per second, which is consistent with experimental observations. The model also predicted a non-linear increase in propulsive efficiency from 0.22 to 0.32 while the required mechanical power quadrupled. The efficiency increase was only minor for wave amplitudes above the experimental reference value, whereas the cost of transport rose significantly.Supplementary material available online at http://jeb.biologists.org/cgi/content/full/215/22/4015/DC1