Summary The randomized response technique was used in a household survey of approximately 2,000 rural and 2,000 urban households in Misamis Oriental Province in the southern Philippines in order to determine the extent of purposive concealment of death. The estimated number of deaths deliberately not revealed to the interviewers was 50 per cent or higher. Adjusted crude death rates of 11.5 and 13.4 per 1,000 population were computed for urban and rural areas, respectively, by adding estimated concealed deaths to deaths reported to the interviewers. Application of stable population techniques and of model life tables suitable to the Philippine setting, while not permitting definite conclusions, provided reasons for believing that these adjusted death rates are close to the true mortality situation in the study areas. Randomized response data further indicate that approximately 75 per cent of urban deaths and 47 per cent of rural deaths of the population studied were not registered with municipal authorities. The authors postulate that failure to register deaths with municipal authorities, together with fear of legal involvement if this failure becomes known outside the immediate neighbourhood, is a major reason for the purposive concealment of death in household surveys.