This paper proposes an information-aware (IA) adaptive data acquisition system for IoT applications. The IA module obtains valid information adaptively by three techniques: 1) clock-free slope detector, 2) signal-dependent full-scale feature, 3) the IA algorithm which combines the features in amplitude and time domains. The IA module adapts both resolution and full scale to the instant valid data while rejects the invalid ones, thus saving its power and the power of the subsequent stages. The event-driven system has 128 nW power consumption, adaptive 3~7 bits resolution, and 15x higher data compression ratio. It precisely fits the emerging ubiquitous IoT applications.