This paper discusses the effect of random errors in power meter readings by the six-port reflectometer. By means of six-port techniques, the determination of the reflection coefficient (Γ) of a divice under test is reduced to the problem of finding a common intersection of three circles in the complex plane. Since the intersection usually forms a cluster due to the effect of measurement error, the extraction of a single value from the cluster including the radical center of the three circles is required. Two types of methods are presented for determining Γ. One uses a linear solution for the radical center, and the other is a statistically based nonlinear solution. In order to improve measurement accuracy, the effect of random errors in the sidearm power meter readings and due to the influence of the q-point locations are investigated for each method. By adding a random variation of 0.5% onto each of the three port power ratios, the variance distributions of Γ over the entire area of the Smith chart are simulated for comparison of these two solutions. The three dimensional variance distribution chart reveals that only the nonlinear solution suffers a variance increase shown as a ridgelike peak along the lines of centers of the three circles. As a result of computer simulations, it is clarified that the reflectometer has the property of measurement accuracy dependence on the value of Γ. A new type of six-port model is suggested, which is unlikely to be affected by random errors in the nonlinear solution.