Radio frequency identification (RFID) is a flexible deployment technology that has been adopted in many applications especially in supply chain management. RFID system used radio waves to perform wireless interaction to detect and read data from the tagged object. However, RFID data streams contain a lot of false positive and duplicate readings. Both types of readings need to be removes to ensure reliability of information produced from the data streams. In this paper, a single approach, which based on Bloom filter was proposed to remove both dirty data from the RFID data streams. The noise and duplicate data filtering algorithm was constructed based on bloom filter. There are two bloom filters in one algorithm where each filter holds function either to remove noise data and to recognize data as correct reading from duplicate data reading. Experimental results show that our proposed approach outperformed other existing approaches in terms of data reliability