A randomized controlled trial was carried out for 1 y in three tertiary and teaching hospitals, in Addis Ababa (Ethiopia), Yogyakarta (Indonesia) and Merida (Mexico), to study the effectiveness, feasibility, acceptability and cost of kangaroo mother care (KMC) when compared to conventional methods of care (CMC). About 29% of 649 low birthweight infants (LBWI; 1000-1999 g) died before eligibility. Of the survivors, 38% were excluded for various reasons, 149 were randomly assigned to KMC (almost exclusive skin-to-skin care after stabilization), and 136 to CMC (warm room or incubator care). There were three deaths in each group and no difference in the incidence of severe disease. Hypothermia was significantly less common in KMC infants in Merida (13.5 vs 31.5 episodes/100 infants/d) and overall (10.8 vs 14.6). Exclusive breastfeeding at discharge was more common in KMC infants in Merida (80% vs 16%) and overall (88% vs 70%). KMC infants had a higher mean daily weight gain (21.3 g vs 17.7 g) and were discharged earlier (13.4 vs 16.3 d after enrolment). KMC was considered feasible and presented advantages over CMC in terms of maintenance of equipment. Mothers expressed a clear preference for KMC and health workers found it safe and convenient. KMC was cheaper than CMC in terms of salaries (US$ 11,788 vs US$ 29,888) and other running costs (US$ 7501 vs US$ 9876). This study confirms that hospital KMC for stabilized LBWI 1000-1999 g is at least as effective and safe as CMC, and shows that it is feasible in different settings, acceptable to mothers of different cultures, and less expensive. Where exclusive breastfeeding is uncommon among LBWI, KMC may bring about an increase in its prevalence and duration, with consequent benefits for health and growth. For hospitals in low-income countries KMC may represent an appropriate use of scarce resources.