High-velocity air-water flows in hydraulic structures, such as spillways and low-level outlets, are characterised by high turbulence levels and strong self-aeration. In air-water flows, the void fraction varies from values close to zero at the invert up to near unity in the upper spray region. Because of the strong aeration, non-intrusive optical measurement techniques can only be applied to a limited extent, and intrusive phase-detection probes are widely used to estimate air-water flow properties, including void fraction, interfacial velocity, and chord sizes. However, these probes are often used without independent validation. Herein, this study systematically investigated the uncertainties of phase-detection probe measurements of individual bubbles traveling at a wide range of velocities (up to 7.5 m/s) in a vertically upward bubbly pipe flow. Velocities and chord times of air bubbles were simultaneously measured with an intrusive dual-tip conductivity probe and stereo-view high-speed videos, and a comparative analysis identified various effects of bubble-probe interactions on the measurement accuracy of the bubble velocities. The bubble deceleration due to these bubble-probe interactions were quantified. For standard interactions (i.e., bubbles undergoing no significant deformation and slowing down), the velocity estimates compared well with a theoretical model (errors less than 2 %), while larger errors were found for non-standard interactions such as those characterised by crawling, drifting, and waking mechanisms. In addition, a novel correction scheme for chord times was proposed, improving the accuracy of chord length, bubble size and void fraction estimations in air-water flows.