The measurement of air temperature, mean air speed, and turbulence intensity is required in order to assess air distribution and draft discomfort in ventilated rooms. The measurements are also used for validation of computational fluid dynamics (CFD) predictions. The uncertainty of the measurements must be known in order to perform reliable assessment and validation. At present, a low-velocity thermal anemometer (LVTA) with an omnidirectional (spherical) sensor is most often used in practice for measuring air speed due to its low price and easy and convenient operation. The accuracy of the speed measurement and of the draft risk determination can be affected by numerous factors, for example, directional sensitivity of the velocity sensor, natural convection flow generated by the heated velocity sensor, dynamic response of the anemometer, calibration of the anemometer, velocity and temperature gradients in the flow of measurement, etc. The impact of these factors can be minimized substantially by improvement of the anemometer design, proper use of the instrument during measurement, and correction of the measured data. However, the extent to which the measuring accuracy can be improved is limited. In this paper, the combined impact of error sources on the accuracy of mean speed, standard deviation of speed, and turbulence intensity that may occur during measurements with LVTAs is analyzed. The minimum uncertainty that is realistically achievable in practice is identified. The requirements for low-velocity anemometers prescribed in the present standards are critically reviewed and revised New requirements that will decrease the uncertainty of low-velocity measurements are suggested for inclusion in future ventilation standards. The uncertainty in determination of draft discomfort is defined. Thus, the definition of realistic requirements in thermal comfort standards as well as validation of CFD predictions is made possible.