One of the most powerful statistical data analysis technique is undoubtfully regression analysis. Apart from the classical regression based on the minimization of the squared errors, the author introduce the interval regression analysis, formulated as a linear programming problem. Three such formulations are proposed together with some calculation examples. Finally, the relation between these interval regression models and a fuzzy liner regression model is described; two subcases are considered whether the data themselves are fuzzy or not. Extensive reference is made to the fuzzy literature, opening a view on several different alternatives including some neural network models.