A general regression model is derived which describes the dependence of the cosmic-ray neutronic component on the air pressure. As the residuals are correlated the ordinary least-squares method is not the best one to estimate the pressure effect. Alternative methods are discussed. Five methods are used to estimate the pressure coefficient at Upsala during the years 1957–1964. It turns out that a modified two-station method gives the best result and also that there exists a solar cycle dependence of the pressure coefficient.