Abstract. Data measured in wireless sensor networks are inherently imprecise, due to a number of reasons, and aggregate queries are often used to analyze the collected data in order to alleviate the impact of such imprecision. In this paper we will deal with the imprecision in the measured values explicitly by employing a probabilistic approach and we focus on one particular type of aggregate query, namely the SUM query. We consider that sensors in the network may, operate (all collectively at the same time) in two different modes: (1) returning a finite set of discrete values with a probability attached to each value, or (2) a continuous probabilistic density function over a possibly infinite set of possible values. Our foremost concern is to present the first algorithms to efficiently compute the probabilistic SUM according to the possible world semantics, i.e., without any loss of information. Furthermore, we show how this query can be efficiently updated in dynamic environments where sensor values change often and we show techniques to distribute computation over all network nodes. Our experimental results show that processing queries in-network and incrementally as opposed to collecting the measured values from all nodes at the base station and computing the answer centrally, can reduce the total number of messages sent by at least 50%, thus saving energy and extending the network's lifetime, a chief concern regarding wireless sensor networks.