The influence of temperature-induced changes in water viscosity on the swimming performance and kinematics of larval Atlantic herring (Clupea harengus) was examined using high-speed video recording. The physical effects of viscosity were measured separately from the physiological (Q10) effects of temperature by increasing the viscosity using methyl cellulose. Voluntary swimming speeds of large larvae (18.2 mm total length) were characterized by Reynolds numbers based on length (ReL) between 100 and 500 and varied with temperature and viscosity. Speeds of small larvae (9.6 mm) at ReL between 25 and 125 were strongly affected by viscosity, but virtually unaffected by temperature at equal viscosities. Speeds of large larvae were modulated by transverse tail speed. Small (viscosity-dominated) larvae altered both transverse tail speed and tail amplitude to vary their swimming speed. Stride lengths for both sizes of larvae followed predictions for viscous-regime swimming until ReL>450. The combined data suggest that the viscous hydrodynamic regime for larval herring extends to at least ReL=300 and that viscosity could be important up to ReL of approximately 450. Because the physical effects of viscosity supplement the physiological effects of temperature on locomotor performance (when ReL is below approximately 300), indices such as Q10 can greatly overestimate the dependence of physiological processes on temperature, as demonstrated by an example.