Estimated of the vertical component of water velocity are made with measurements of the pressure-change rate and rotation rate recorded by the free-fall vehicle “Cartesian diver” (CD). Using buoyancy control, this device moves alternately up and down, essentially at its terminal velocity through the surrounding water. Four angled wings fix the lift and rotational characteristics of the instrument. The rate of rotation is directly proportional to the terminal velocity of the instrument and is not sensitive to vertical water accelerations except at scale lengths somewhat less than the instrument height. When the pressure distribution is hydrostatic, the pressure-change rate is proportional to the sum of the instrument's terminal velocity and the vertical water velocity. Because the terminal velocity is a function of the instrument buoyancy, which varies slowly with depth, the fluctuations in vertical water velocity are well resolved by the fluctuations of the pressure-change-rate sensor. The orbital velocity of barotropic surface waves does not contribute to these measurements because isobars remain very nearly fixed to water particles. The vertical velocity of baroclinic flows, whether associated with internal waves, turbulence, or convective motions, is detectable in principle. Data from repetitive profiles over the depth range from 100 to 350 in are used to estimate the vertical components of velocities with a resolution of 0.1 mm s−1. It is shown that at high frequencies, in a band centered near 0.1 Hz, a pressure disturbance caused by the nonlinear interaction of opposed surface wave trains is sometimes prominent and essentially limits the detection of short vertical wavelength internal motions to 2 m (the maximum vertical distance traveled in 10 s). In open-sea observations in both the Atlantic and Pacific, root-mean-square (rms) vertical velocities of 3–4 cm s−1 at vertical scales of 20 m and greater are commonly observed along with fluctuations of 1–10 mm s−1 with vertical scales down to 2 m.