Concentrations of carbon dioxide, methane and nitrous oxide were continuously measured over several weeks at a range of positions within (and outside) each of a fattening piggery, a broiler house and a dairy cow house, using an automated gas chromatograph. For the piggery and the broiler house, both of which were force-ventilated, the ventilation rate was measured using fan-wheel anemometers. For the cow house, which was naturally-ventilated via space boarding and a roof ridge vent, the ventilation rate was measured using a novel carbon monoxide tracer gas technique. Emission rates were determined from the products of ventilation rates and the concentration differences across the buildings. The rates of carbon dioxide emission measured were compared with those predicted from literature on metabolic rates of production. Agreement was within 15% for the force-ventilated buildings and 20% for the naturally-ventilated building, giving additional validation of the 2 methods of measuring ventilation rate. For the broiler house, no significant emissions of either methane or nitrous oxide were found: it is probable that the very dry litter in this house had inhibited the microbial processes necessary to produce both gases. The cow house gave, per 500 kg liveweight housed, greater emissions of both methane and nitrous oxide than did the piggery (4 times as much for methane; twice as much for nitrous oxide). All methane from the dairy cow house was from the cows themselves: none was from the slurry. It is concluded that the best option for reducing overall greenhouse gas emissions from housed livestock is the manipulation of bovine diets.