This paper describes the purpose and methods of a single-blinded, clustered and randomised trial of the health impacts of insulating existing houses. The key research question was whether this intervention increased the indoor temperature and lowered the relative humidity, energy consumption and mould growth in the houses, as well as improved the health and well-being of the occupants and thereby lowered their utilisation of health care. Households in which at least one person had symptoms of respiratory disease were recruited from seven predominantly low-income communities in New Zealand. These households were then randomised within communities to receive retrofitted insulation either during or after the study. Measures at baseline (2001) and follow-up (2002) included subjective measures of health, comfort and well-being and objective measures of house condition, temperature, relative humidity, mould (speciation and mass), endotoxin, beta glucans, house dust mite allergens, general practitioner and hospital visits, and energy or fuel usage. All measurements referred to the three coldest winter months, June, July and August. From the 1352 households that were initially recruited, baseline information was obtained from 1310 households and 4413 people. At follow-up, 3312 people and 1110 households remained, an 84% household retention rate and a 75% individual retention rate. Final outcome results will be reported in a subsequent paper. The study showed that large trials of complex environmental interventions can be conducted in a robust manner with high participation rates. Critical success factors are effective community involvement and an intervention that is valued by the participants.