The degree to which local health systems contribute to reductions in under-five mortality in severely impoverished settings has not been well documented. The current study compares the under-five mortality in the Hôpital Albert Schweitzer (HAS) Primary Health Care Service Area with that for Haiti in general. HAS provides an integrated system of community-based primary health care services, hospital care and community development. A sample of 10% of the women of reproductive age in the HAS service area was interviewed, and 2390 live births and 149 child deaths were documented for the period 1995-99. Under-five mortality rates were computed and compared with rates for Haiti. In addition, available data regarding inputs, processes and outputs for the HAS service area and for Haiti were assembled and compared. Under-five mortality was 58% less in the HAS service area, and mortality for children 12-59 months of age was 76% less. These results were achieved with an input of fewer physicians and hospital beds per capita than is available for Haiti nationwide, but with twice as many graduate nurses and auxiliary nurses per capita than are available nationwide, and with three cadres of health workers that do not exist nationwide: Physician Extenders, Health Agents and Community Health Volunteers. The population coverage of targeted child survival services was generally 1.5-2 times higher in the HAS service area than in rural Haiti. These findings support the conclusion that a well-developed system of primary health care, with outreach services to the household level, integrated with hospital referral care and community development programmes, can make a strong contribution to reducing infant and child mortality in severely impoverished settings.