According to tidal friction calculations in which estimates of the current dissipation factor are used, the moon was close to the earth less than two billion years ago. The most plausible solution to this ‘time scale’ problem appears to be a much lesser extent of shallow seas in the past; a coupling of the moon's orbit with Venus is also a possibility. The same calculations indicate that the moon had an inclination of at least 10° to the equator, and that the earth's rotation period was five hours. Fission hypotheses, of which the most carefully developed is by O'Keefe, have not yet dealt successfully with these constraints. Fission theories are also unsatisfyingly episodic: matter must first lose angular momentum to fall into the earth, and then regain it to separate. All capture hypotheses are inherently improbable, particularly if the moon came from a different enough part of the solar system to account for the moon's density difference. The least implausible capture hypothesis, by Urey and MacDonald, entails collision with a pre‐existing earth satellite at 30–40 earth radii, but the reasoning behind it is not compelling. Most compatible with the idea that formation of the moon is closely connected to the formation of the earth are hypotheses that the moon formed out of a geocentric swarm of matter to which matter from heliocentric orbit was continually added. The theory of Ruskol seems to incorporate most of the dynamical essentials. However, the model needs to be developed further to include gas drag and other effects of the severe heating associated with chemical fractionation between the earth and moon, as well as to account for the inclination of the lunar orbit. A possibly severe constraint on the distance from the earth at which the moon's formation was completed is a systematic ellipticity of lunar craters pointed out by Öpik; it should be re‐examined after oriented metric photographs from Apollo 15–17 flights