||The emergence of wireless sensor networks (WSNs) has enabled a wide variety of applications, such as environmental monitoring, assisted living, and industrial control. In the meanwhile, handheld devices, especially PDAs (Personal Digital Assistants), are becoming more and more powerful and popular. Therefore, we envision that in some applications, handheld devices will be commonly used to query deployed WSNs. However, there has been little work on developing a lightweight query processor for PDAs to query sensor networks. To address this problem, we propose WinyDB, a relational query processing system on Windows-CE based PDAs for sensor networks. One of the main features of WinyDB is that multiple PDAs running WinyDB can answer queries collaboratively. This collaborative query processing feature is useful in that it improves both the power efficiency and the data quality. We have proposed collaborative query processing techniques to handle selection, projection, aggregation, and window join queries. Furthermore, we have implemented the entire WinyDB system on Windows-CE based PDAs. Our experiments with the system running on multiple PDAs querying simulated sensor networks have shown promising results on collaborative query processing.