RFID systems have been widely deployed for various applications such as supply chain management, indoor localization, inventory control, and access control. This paper deals with the fundamental problem of estimating the number of arriving and departing tags between any two time instants in dynamically changing RFID tag populations, which is needed in many applications such as warehouse monitoring and privacy sensitive RFID systems. In this paper, we propose a dynamic tag estimation scheme, namely DTE, that can achieve arbitrarily high required reliability, is compliant with the C1G2 standard, and works in single as well as multiple-reader environment. DTE uses the standardized frame slotted Aloha protocol and utilizes the number of slots that change their values in corresponding Aloha frames at the two time instants to estimate the number of arriving and departing tags. It is easy to deploy because it neither requires modification to tags nor to the communication protocol between tags and readers. We have extensively evaluated and compared DTE with the only prior scheme, ZDE, that can estimate the number of arriving and departing tags. Unfortunately, ZDE can not achieve arbitrarily high required reliability. In contrast, our proposed scheme always achieves the required reliability. For example, for a tag population containing 104 tags, a required reliability of 95%, and a required confidence interval of 5%, DTE takes 5.12 seconds to achieve the required reliability whereas ZDE achieves a reliability of only 66% in the same amount of time.