The purpose of this study was to explore the reliability of estimating three-dimensional (3D) angular kinematics and kinetics of a swimmer derived from digitized video. Two high-level front crawl swimmers and one high level backstroke swimmer were recorded by four underwater and two above water video cameras. One of the front crawl swimmers was digitized at 50 fields per second with a window for smoothing by a 4^th order Butterworth digital filter extending 10 fields beyond the start and finish of the stroke cycle (FC1), while the other front crawl (FC2) and backstroke (BS) swimmer were digitized at 25 frames per second with the window extending five frames beyond the start and finish of the stroke cycle. Each camera view of one stroke cycle was digitized five times yielding five independent 3D data sets from which whole body centre of mass (CM) yaw, pitch, roll, and torques were derived together with wrist and ankle moment arms with respect to an inertial reference system with origin at the CM. Coefficients of repeatability ranging from r = 0.93 to r = 0.99 indicated that both digitising sampling rates and extrapolation methods are sufficiently reliable to identify real differences in net torque production. This will enable the sources of rotations about the three axes to be explained in future research. Errors in angular kinematics and displacements of the wrist and ankles relative to range of motion were small for all but the ankles in the X (swimming) direction for FC2 who had a very vigorous kick. To avoid large errors when digitising the ankles of swimmers with vigorous kicks it is recommended that a marker on the shank could be used to calculate the ankle position based on the known displacements between knee, shank, and ankle markers.