The design of broadband and high frequency circuits requires accurate information on the complex dielectric permittivity of low loss printed circuit board materials. The assumption that the relative permittivity is purely real and frequency-independent leads to physically-inconsistent simulation results and a poor prediction of actual system performance. The relative permittivity is obtained from measurements on two different-length microstrip lines . The effective permittivity is then converted to relative permittivity. Genetic algorithm optimization is used to determine the relaxations and relative weights of the relaxation frequencies of a multi-term Debye model in order to achieve a good match with the measured results. A close look at the Debye model coefficients leads one to consider ways to reduce the terms in the model in order to simplify the model and decrease overall simulation time. This paper investigates how the number of terms used in a multi-term Debye material model impact the accuracy of the computed complex propagation constant in low loss printed circuit boards.