In 2012, National Institute of statistics of Rwanda conducted the 4th population and housing census
whereby the population projection was done about fertility and mortality .Based on the assessment of past
mortality, it is assumed that life expectancy at birth in Rwanda will increase linearly up to 70 years with of course
a significant decrease in infant mortality and under five mortality rates by the end of the projection period
2032.Therefore,the forecasts results should be inaccurate as long as the forecasting period becomes too big.
Objective of this study was to apply Leslie matrix, Lee carter and ARIMA models in forecasting population
dynamics in Rwanda. Leslie matrix model was used to determine the annual growth rate of female population
which was r =2.33% .By ARIMA models, Forecasts showed a decreasing trend in IMR and U5MR from 2012 to
2015. The results showed that models used had a higher predictive ability because of their small root mean square
errors where RMSE was found to be 0.124456 for IMR and 0.068576 for under 5 mortality rates (‰).