The objective of this work is to model the extreme temperature climatic records of Villa Clara Cuba and see if there is a trend in them, in addition the variable date on which they occurred was modeled, with the help of the Regressive Objective Regression (ROR). A database from 1966 to 2020 of the 4 weather stations with the account of the province of Villa Clara is used. The explained variance of the models is 100% for the maximum temperature and 99.8 for the minimum with errors of 0.58 and 1.4ºC. You can estimate the graphs for the maximum temperature as for the minimum with the predicted values ​​and the errors that the model commits. The trend for the date of the maximum trend is negative while for the minimum it is positive. The records depend on the temperature returned in 1 month (LAG1T) and the temperature returned in 12 months (LAG12T), both for the maximum TX and for the minimum TN, as well as the station value. The correlations between the actual and predicted value for the maximum and minimum temperature records and for the date models are high, greater than 90% and 99% variable.