Local bubble size distributions (BSD) were measured 3 at multiple locations and agitation conditions in a Rushton turbine agitated 14 dm(3) vessel. Experiments were made with air-deionized water and CO2-n-butanol systems. Three commonly used measurement techniques, capillary suction probe (CSP), phase Doppler anemometry (PDA) and digital imaging (DI), were tested, allowing critical evaluation of measurement techniques. All techniques revealed qualitatively reasonable variation of BSDs with the measurement point and agitation conditions. The quantitative comparison of techniques, instead, revealed a notable deviation. The main reason was related to the limited detectable bubble size range of each technique. The DI detected 0.1-8.0 mm bubbles, PDA 0.005-1.4 mm bubbles and CSP 0.4-8.0 mm bubbles. The sampling errors in the CSP, the identification errors of bubbles from the images in the DI and the optical distortions in the PDA were other reasons for the deviations. It appeared that the shape of number BSDs and arithmetic mean bubble sizes were captured better with the PDA than with DI or CSP. The DI and CSP techniques instead detected the volumetric BSDs and Sauter mean bubble sizes better than PDA. The comparison indicates that the limitations of measurement techniques are a major reason for the spread between bubble size studies for stirred tanks.