The study focuses on Modelling and Forecasting Infant Mortality using Time Series methodology called ARMA. The data used for the study was obtained from the hospital record of the General Hospital, Asubiaro, Osun State, south western, Nigeria. The data was subjected to various stationarity tests (graphical and unit root test) and the data was stationary at the second difference. Thereafter, various ARMA model was fitted from which ARMA(1,1) was chosen. The model fitted has a very powerful forecasting ability as Theil-U index value of (0.722313) obtained was moderate, bias proportion (0.0076) and variance proportion (0.00342) almost tends to zero, and covariance proportion (0.99608) is very high. All these combined together enhanced good forecast performance.