Uncertainty measures for the discrete data system can not be effective for the measurement of uncertainty ofneighborhood system. Therefore, the neighborhood rough set model is introduced to propose measure approaches such as neighborhood accuracy, neighborhood knowledge granularity as well as neighborhood approximation accuracy based on knowledge granularity. The effectiveness of the approaches are verified theoretically. Experimental results show that approximation accuracy based on knowledge granularity with stricter monotonicity outperforms the neighborhood approximation accuracy for the uncertainty measurement of the neighborhood system.