The formalism of the exact DSN method, also called the I* method, is recalled. A fully analytical integration is proposed to compute the three-dimensional array I*(k, l, m), which is the crucial point of such a theory. Comparisons with the numerical quadratures proposed so far show that a significant CPU time savings is achieved (a factor of 1000 for S8), and, moreover, a few symmetry properties relax the storage requirements by a factor of ∼4.