Bighead, silver, and grass carps are invasive in the waterways of central North America, and grass carp reproduction in tributaries of the Great Lakes has now been documented. Questions about recruitment potential motivate a need for accurate models of egg and larval dispersal. Quantitative data on swimming behaviors and capabilities during early ontogeny are needed to improve these dispersal models. We measured ontogenetic changes in routine and maximum swimming speeds of bighead, grass, and silver carp larvae. Daily measurements of routine swimming speed were taken for two weeks post-hatch using a still camera and the LARVEL program, a custom image-analysis software. Larval swimming speed was calculated using larval locations in subsequent image frames and time between images. Using an endurance chamber, we determined the maximum swimming speed of larvae (post-gas bladder inflation) for four to eight weeks post-hatch. For all species, larval swimming speeds showed similar trends with respect to ontogeny: increases in maximum speed, and decreases in routine speed. Maximum speeds of bighead and grass carp larvae were similar and generally faster than silver carp larvae. Routine swimming speeds of all larvae were highest before gas bladder inflation, most likely because gas bladder inflation allowed the fish to maintain position without swimming. Downward vertical velocities of pre-gas bladder inflation fish were faster than upward velocities. Among the three species, grass carp larvae had the highest swimming speeds in the pre-gas bladder inflation period, and the lowest speeds in the post-gas bladder inflation period. Knowledge of swimming capability of these species, along with hydraulic characteristics of a river, enables further refinement of models of embryonic and larval drift.