Modeling solar radiation is a necessity for the utilization of the benefits it brings to mankind. Time series analysis has proved to stand out amidst other statistical tools when estimating and forecasting solar radiations and their variations. In this paper, a mixture of the Autoregressive Moving Average (ARMA) and Generalized Autoregressive Conditional Heteroscedasticity (GARCH) time series models were implemented on the solar radiation series for three (3) representative meteorological stations in Nigeria namely; Ibadan, Sokoto and Port Harcourt to capture and model the conditional mean and volatility that may exist in the series. After subjecting the models to some evaluation metrics for model adequacy, the results gave appropriate ARMA models for the stations and indicated the presence of volatility in the radiations series. Furthermore, a-week-ahead forecasts were conducted for these stations using the ARMA-GARCH model combination which gave close convergence with the actual radiations for year 2016.