The observation that gas-vacuoles of blue-green algae do not disappear under vacuum has given rise to the belief that they are gas-tight structures. In this paper the following evidence is presented for the gas-vacuole membrane being freely permeable to gases. (1) In an attempt to identify the vacuole gas by mass spectrometry it was found that, after thorough evacuation, little or no gas could be obtained from the gas-vacuoles. (2) Using an ultrasonic method of destroying gas-vacuoles in conjunction with a mano-metric method of measuring the amount of gas released, it was demonstrated that the pressure of the vacuole gas was usually at about one atmosphere, but that it fell if the pressure over the algal suspension was decreased. (3) Modifications were made to a Warburg apparatus which enabled the gas exchange capacity of a liquid to be measured, and the apparatus was used to show that, when the pressure over an algal suspension was raised and lowered, gas passed in and out of the gas-vacuoles. The amount of gas they exchanged was directly proportional to the pressure change, and this provided a new means of estimating gas-vacuole volume. The gas-vacuole membranes appeared to be very permeable to nitrogen, oxygen and argon. (4) If the gas pressure over an algal suspension is raised suddenly by about 2 atmospheres, the gas-vacuoles collapse. But if the pressure of the gas is raised slowly, it dissolves in the suspending medium and diffuses into the gas-vacuoles, preventing the differential pressure required to deflate them from being established across the membrane. These findings make it necessary to reconsider both the functions gas-vacuoles could fulfil and the way in which they might form. They must be rigid structures and it is proposed that they may be self-erecting. Mechanisms preventing the accumulation of water in the gas-vacuole are discussed.