Amongst different kinds of frosts, radiation frost is very important; because of frequency occurrence and is controllable well with active protection methods in agricultural sector. Along with radiation frost monitoring, Precise forecasting of the minimum temperature and hourly estimation of its variations (trend) during the nights with frost event for starting and ending time determination of the active protection methods is satisfied. Therefore, using an experimental forecasting model, which can calibrate to meet local conditions and have a simple application, it seems essential. Thereafter, this study aimed at predicting the minimum temperature using a simple regression model and estimating the temperature trend during these kinds of nights. In this paper, using hourly synoptic data of Mashhad station for January, February and March during 1999-2010, meanwhile separating radiation frost from advection frost, dew point temperature and dry temperature two hours after sunset embedded for developing the prediction model of minimum temperature. Then, according to the predicted minimum temperature, the temperature trend during the night hours with radiation frost event was incident prediction. In the significant confidence level of 99 percent, Pearson correlation coefficient value between observed and predicted minimum temperatures based on developed model was 0.93. The amount of root mean square error (RMSE) for the developed model is 1.63oC and the amount allowed this model predicted 76 percent. In forecasting the occurrence of temperatures during the nights with radiation frost event in 2010, the amount of RMSE is 0.7oC and correlation coefficient is more than 0.90 (significant at 0.01) respectively.