The surfaces of copper foils used in Printed circuit boards are normally roughened, for better adhesion to the boards. However, as the speed of system busses increase at a rate predicted by Moore’s law, the negative impact of surface roughness on power losses has become more important. The empirical Hammerstad equation, which has long been used to quantify surface roughness power losses, has failed to correlate with measurements. At low and high frequencies, the Hammerstad equation tends to underestimate or overestimate loss depending on the copper profile. As a result, accurate methods of predicting power loss due to surface roughness need to be developed. Current numerical methods using 3-D field solvers are prone to errors due to difficulties in meshing dimensions in the order of 2-11 microns and inaccurate estimate of the coupling between the port and the surface disturbance. A 3D model of surface roughness effects at different frequencies using analytical methods is compared to numerical methods using 3-D field solvers in this dissertation. In addition it is shown that the main contributing factor in power loss at high frequencies is due to micrometer size “snowballs”.