Important constraints on possible molecular mechanisms of general anaesthesia are derived from a quantitative reappraisal of data on the potency of general anaesthetics on whole animals. Despite their popularity, theories that invoke lipids as the prime target do not look at all promising, and available data point much more plausibly to a direct effect on particularly sensitive proteins. Structural changes of proteins on binding general anaesthetics are probably small but may be sufficient to perturb normal function; alternatively, anaesthetics may compete with an endogenous ligand. The phenomenon of pressure reversal of anaesthesia may simply be due to anaesthetics being squeezed away from their target sites.