The aim of this study is to find an efficient and robust model for building temperature information systems in South-East Nigeria. The study obtained daily mean temperature data records for a period of 10years of the capture cities of Enugu, Abakaliki and Owerri, and applied the data to several forecasting models: 3 & 4 point moving averages (MA), the Single Exponential Smoothing (SES) and the time dependent regression model for intercept and non-intercept models as well as linear and non-linear models. The comparison of various forecasting models was made based on the following performance evaluation methods: F-values, Mean Square Error (MSE) and Root Mean Square Error (RMSE) where applicable. The findings show that the power model with statistical characteristics of F-values = 1513.71(Enugu), 1523.622(Abakaliki) and 1514.103(Owerri), MSE = 0.655(Enugu), 0.6495(Abakaliki), and 0.5925(Owerri), and RMSE = 0.80524(Enugu), 0.80292(Abakaliki) and 0.76703(Owerri), is the best model for temperature information systems because of its consistency in minimizing errors, and largeness of F-values. This is followed by the single exponential smoothing technique and logarithmic model. This study therefore presents and recommends the power regression model as the most robust model for temperature forecasting in South-East Nigeria.