This study examines infant mortality, a critical global health issue, particularly in developing countries with economic disparities affecting healthcare access. It compares the forecasting accuracy of three econometric models Vector Error Correction Model (VECM), Vector Autoregressive (VAR), and Bayesian VAR (BVAR), to predict infant mortality rates (IMR). The analysis uses data on IMR, neonatal mortality rates (NMR), and Ugandan GDP and GDP per capita (GDPP), from 1954 to 2016, assessing model performance through statistical measures like Mean Squared Error (MSE) and Theil’s U-statistic. The results reveal significant long-term relationships between IMR, NMR, GDP, and GDPP, with VECM being the most accurate model for long-term forecasting, achieving an adjusted R-squared of 97.7%. Impulse response analysis shows GDP positively impacts IMR, while GDPP has a stronger long-term effect in reducing IMR. For NMR, GDP has a negative effect, while GDPP shows a positive response over time. Granger causality tests confirm bidirectional causality between GDPP and IMR, while IMR unidirectionally influences GDP. Projections indicate Uganda’s IMR could decline to 17 deaths per 1,000 live births by 2035, although the decline in NMR will slow. In conclusion, short-term forecasts are best modeled by ARIMA, while long-term forecasts are more accurately captured by VECM. GDPP has a more substantial impact on reducing IMR and NMR than GDP, highlighting the importance of equitable resource distribution and macroeconomic growth for improving child survival rates.