Severe acute malnutrition (SAM) causes 1 million deaths in children annually by making children susceptible to common infections. The World Health Organisation (WHO) recommends that children with SAM should receive antibiotics together with nutritional rehabilitation. Children with SAM and complications including signs of infection or severe metabolic disturbance are referred for hospital admission. However, most admissions with SAM present directly to hospital because of severe illness, and their SAM is only detected during clinical assessment. At the four hospital sites for the proposed trial, between 15% and 18% of paediatric admissions between the ages of 2 and 59 months have SAM. In sub-Saharan Africa, up to 30% of children admitted to hospital with complicated SAM die, usually from severe infection. Mortality is highest amongst those who also have HIV infection. There are reports that bacteria isolated from children with SAM, when tested in the laboratory, are often not susceptible to the recommended first-line antibiotics. In Kenya we have conducted long term surveillance of bacterial infections. Amongst children with SAM, non-susceptibility has risen: in the last 5 years, more than one third of bacteria isolated at admission to hospital are non-susceptible to the recommended antibiotics. Because children with SAM are vulnerable to infection, this to result in death rather than simply a prolonged hospital stay. An alternative antibiotic, ceftriaxone, is cheaper the currently recommended combination and only has to be given once a day instead of four times. Much less resistance to ceftriaxone is reported. Ceftriaxone would be used as first line treatment if such a child were admitted to hospital in the UK. At first sight, it seems that ceftriaxone would be a more appropriate antibiotic, and it could reduce deaths. However, a significant concern is that ceftriaxone is known to rapidly induce resistance to multiple classes of antibiotics. This could mean that subsequent infections could be harder, and more expensive, to treat. Furthermore, studies have not shown a clear relationship between laboratory susceptibility testing and actual outcomes. In determining policy for empiric antimicrobials for this vulnerable population, potential benefits of reduced mortality, quicker recovery and reduced costs must be weighed against potential risks of infections that are difficult and expensive to treat. There is currently no evidence to inform this decision. A second question in the antibiotic treatment of SAM is the value of metronidazole. Current WHO guidelines suggest that metronidazole may be optionally used although it has never been tested in a clinical trial. It is effective against bacteria that cause abnormal overgrowth in the small bowel, and against gut parasites such as Giardia. These conditions are common amongst children with SAM and may cause malabsorption of nutrients and diarrhoea. Treating them improve nutritional recovery. Results of small studies suggest this may be the case. However, metronidazole can cause nausea, vomiting and other toxicities which could impede nutritional recovery. We propose an efficiently designed trial to test both ceftriaxone and of metronidazole against standard care for the outcomes of mortality and nutritional recovery. First we will determine the optimal dosing for the drugs in malnourished children. We will carefully investigate children for infections and the antibiotic susceptibility of bacteria isolated determined. An economic analysis will measure the cost-benefit ratio of each strategy and overall costs of treatment for SAM. The trial will be run at 2 rural and 2 urban hospitals in Kenya. The results are expected to have direct impact on antibiotic policy for the management of SAM in hospitals in Africa and will provide unique information that will contribute to global efforts to combat the threat of antimicrobial resistance.