Background: Kangaroo care was introduced in Bogota, Colombia in 1979 by Dr Edgar Ray and Dr Hector Martinez due to the shortage of resources and the large number of premature babies that needed special care. Kangaroo care implies direct skin-to-skin contact between the mother and her premature/newborn baby. The advantages of kangaroo care are well known and have been published widely. In the National District Hospital in Bloemfontein, South Africa, stable babies are admitted from 1.2 kg and above and the babies are discharged at around 1.8 kg, when they are able to drink adequately. Babies above 1.2 kg are given kangaroo care by their mothers in the day, but sleep in an incubator at night. Babies with a weight of between 1.5 kg and 1.6 kg are started with around-the-clock kangaroo care. The aim of this study was to determine predictors for good and poor outcomes in kangaroo mother care practiced at the primary healthcare level. This information was then used to compile a protocol for kangaroo care in hospitals and maternity sections at the primary care level. The aim of the protocol is to provide specific inclusion and exclusion criteria for kangaroo care, to indicate all the absolute and relative needs for such a unit, to provide guidelines for managing babies with insufficient weight gain, and to provide guidance for follow-up. Methods : In this cohort study, the patient files of premature babies that received kangaroo care in the National District Hospital were evaluated in order to establish indicators for good and poor outcomes. Patient files were selected consecutively from the last entry in the admissions register from May 2005 backwards until June 2003. Data was collected on a standard data-collection form. Reasons for not gaining weight and the need for special investigations were investigated and noted. Poor weight gain was regarded as weight gain of less than 17 g/kg/day. Results : A total of 200 files were audited. In 62% of the cases (95% CI 55.1%; 68.4%), the babies gained weight satisfactorily, while weight gain was unsatisfactory in 38% of the cases. The mean admission weight of the babies was 1 545 g (range 1 100 g to 2 100 g) and the mean discharge weight was 1 800 g (range 1 700 g to 2 100 g). The chance of weight gain was reduced if one of the following occurred: anaemia, low body temperature, inappropriate amount and route of milk administered, sepsis, transport, procedures and other medical conditions. The reinsertion of nasogastric tubes (53%), improved temperature control with improved kangaroo care technique (79%), the correction of anaemia with blood transfusion (12%) and the correction of the volume of milk (5%) were the major corrective steps taken to address the problem. In 29% of the cases, extra energy in the form of FM 85 was added to the breast milk. The type of milk that the babies received, namely breast milk (n=113), premature milk formula (n=40) and a combination of breast milk and formula milk (n=46), did not significantly influence weight gain. Gender, birth weight and gestational age at birth also did not have a significant influence on weight gain. If the problem was properly addressed, the babies started gaining weight within an average of four days. Conclusion : It is possible to render safe kangaroo care to all stable premature babies in a primary healthcare setting, provided that the set guidelines are adhered to. For full text, click here: South African Family Practice 2006;48(9):16-16c