Abstract: In the research of yarn unevenness, the fiber alignment in the yarn is the key to description of the problem. If an assumption is given that the ends of the fibers in yarn follow a uniform distribution along the yarn length, for the given fiber length distribution, then the ideal yarn model can be established geometrically. The Monte Carlo method is used to simulate the variation of the number of fibers in the yarn cross section. In the geometric model, a segment of 8 mm is set to simulate the yarn unevenness in the capacitive type yarn unevenness tester. The comparison shows that under the assumption that fibers in yarn are arranged randomly, the simulated CV of the yarn will increase with the increase of yarn count. The number of fibers in the yarn cross section follows normal distribution asymptotically. The CV of the number of fibers in yarn cross section and the quantity of fibers in the segment are not relevant significantly. The test data illustrates the validity of the model.