Objectives: To evaluate the prognostic value of selected laboratory and growth markers on the short-term risk of mortality in untreated HIV-infected children in resource-limited settings. Design: A meta-analysis of individual longitudinal data on children aged 12 months onwards from 10 studies (nine African, one Brazilian in the 3Cs4kids collaboration). Methods: The risk of death within 12 months based on age and the most recent measurements of laboratory and growth markers was estimated using Poisson regression models, adjusted for cotrimoxazole prophylaxis use and study effects. Results: A total of 2510 children contributed 357 deaths during 3769 child-years-at-risk, with 81% follow-up occurring after start of cotrimoxazole. At first measurement, median age was 4.0 years (interquartile range, 2.2-7.0 years), median CD4% was 15% and weight-for-age z-score -1.9. CD4% and CD4 cell count were the strongest predictors of mortality, followed by weight-for-age and haemoglobin. After adjusting for these markers, the effects of total lymphocyte count and BMI-for-age were relatively small. Young children who were both severely malnourished and anaemic had high mortality regardless of CD4 values, particularly those aged 1-2 years. By contrast, high CD4% or CD4 cell count values predicted low mortality level amongst either children older than 5 years or those younger with neither severe malnutrition nor anaemia. Conclusions: CD4 measurements are the most important indicator of mortality and wider access to affordable tests is needed in resource-limited settings. Evaluation of antiretroviral initiation in children also needs to consider weight-for-age and haemoglobin. Prevention and treatment of malnutrition and anaemia is integral to HIV paediatric care and could improve survival. (c) 2008 Wolters Kluwer Health.