To describe trends of childhood stunting among under-5s in Uganda and to assess the impact of maternal education, wealth and residence on stunting. Serial and pooled cross-sectional analyses of data from Uganda Demographic and Health Surveys (UDHS) of 1995, 2001, 2006 and 2011. Prevalence of stunting and mean height-for-age Z-score were computed by maternal education, wealth index, region and other sociodemographic characteristics. Multivariable logistic and linear regression models were fitted to survey-specific and pooled data to estimate independent associations between covariates and stunting or Z-score. Sampling weights were applied in all analyses. Uganda. Children aged <5 years. Weighted sample size was 14 747 children. Stunting prevalence decreased from 44·8% in 1995 to 33·2% in 2011. UDHS reported stunting as 38% in 1995, underestimating the decline because of transitioning from National Center for Health Statistics/Centers for Disease Control and Prevention standards to WHO standards. Nevertheless, one in three Ugandan children was still stunted by 2011. South Western, Mid Western, Kampala and East Central regions had highest odds of stunting. Being born in a poor or middle-income household, of a teen mother, without secondary education were associated with stunting. Other persistent stunting predictors included small birth size, male gender and age 2-3 years. Sustained decrease in stunting suggests that child nutrition interventions have been successful; however, current prevalence does not meet Millennium Development Goals. Stunting remains a public health concern and must be addressed. Customizing established measures such as female education and wealth creation while targeting the most vulnerable groups may further reduce childhood stunting.