<p>This study develops a structural dynamic response and unknown excitation estimation algorithm that considers multi-rate (MR) sensor data fusion. Nearly all the existing joint input and state estimation algorithms have the underlying assumption that all sensor data are recorded at the same frequency. This limitation impedes algorithms’ application to real structural health monitoring (SHM) projects, as different sensors may have various sampling rates, especially for smartphone-assisted vision-based monitoring. Sensor data loss is also frequently encountered in SHM applications. To solve these problems, a state-of-the-art MR unified linear input and state estimator (MR ULISE) is developed. The measurement update is conducted at each time point using the available sensor data. If sensor data loss is encountered, only the time update is performed. The recently presented Rauch, Tung, and Striebel (RTS) smoothing technique under unknown input is also introduced into the response reconstruction field. The data collected by the numerical and experimental studies provide convincing evidence that the MR ULISE with RTS smoothing algorithm is an efficient and reliable technique under diverse application scenarios, including different structural configurations (e.g., frame and beam), excitation types (e.g., force input and ground motion), and sampling rates, that can obtain satisfactory estimation results for unknown excitation and structural responses, even under fairly low-frequency displacement or strain measurements. This study also facilitates the emerging smartphone-assisted monitoring in civil structures.</p>