An emission intensity protocol to govern long-term international greenhouse gas emission reduction is proposed. The protocol may also be interpreted as a technology protocol. The protocol consists of three parameters: a graduation income, below which countries have no emission reduction obligations; a convergence rate, at which emission intensities should approach that of the most carbon-extensive countries; and an acceleration rate, at which the most carbon-extensive countries should improve their technology over and above the business-asusual scenario. Depending on the parameter values, emission reduction ranges from draconian to almost nil. The graduation income and acceleration rate have the expected effects. The effect of the convergence rate is strongly scenario-dependent; some scenarios, perhaps unrealistically, assume strong technological convergence in the nopolicy case; in other scenarios, adopting ‘best commercial technology in the whole world’ would lead to substantial emission reduction. Not surprisingly, different regions prefer different parameters in the emission intensity protocol. Adopting the opinion of the median voter, atmospheric concentrations of carbon dioxide in the year 2200 would be reduced from 1650 to 950 ppm. This reduction is relatively robust to changes in crucial model parameters. The costs of complying with the emission intensity protocol can be reduced substantially through international trade in emission permits and, in particular, by banking and borrowing.