Noise from many people speaking in eating establishments and other social gatherings is a well known and wide spread problem. The problem is particularly difficult to handle because the sound sources are individual and dynamic, i.e. the speech level increases when the ambient noise level goes up. However, a simple prediction model has been derived that allows estimating the ambient noise due to speech from a large group of people, the main uncertainty being the so-called group size, i.e. the average number of people per speaking person. As a measure of the acoustical quality is suggested the average signal-to-noise ratio when listening to a person speaking to you in a distance of 1 m and the ambient noise is that from other people speaking in the room. The Acoustical Capacity is defined as the number of people that would create a signal-to-noise ratio of -3 dB, which is considered the lower limit for “sufficient” quality of verbal communication under certain preconditions. The Acoustical Capacity is calculated from volume and reverberation time by a very simple equation. The acoustical quality of an eating establishment may be characterized by the ratio between the Acoustical Capacity and the total capacity.