So far, only large-sample properties of the sampling distribution of the intercept term in two-variable regression are known. In this paper we shall derive the exact form of the sampling distribution of the intercept term for the case when the population is bivariate normal. We shall also find the moments of the distribution of the intercept term. We shall show that the distribution is symmetric for all sample sizes, and that the distribution tends towards normality as the sample size increases indefinitely.