Abstract: Based on a lot of network performance management experience, the network performance data’s changing rules and the shortcomings of the equal-interval sample algorithm are studied. The equal-interval sample algorithm is very popular in the present network management systems. A self-adaptive-interval sample algorithm is proposed. This algorithm adjusts the intervals based on the network performance data’s change, and in this way it spends the network bandwidth more reasonably. Theoretical analysis and simulation shows that this algorithm can overcome the ignorance of the performance data’s change of the equal-interval sample algorithm. Spending equal network bandwidth, a better precision can be gained by using the self-adaptive-interval sample algorithm.