In many computer-based applications, temporal information has to be stored, retrieved, and related to other temporal information. Several time models have been proposed to manage temporal knowledge in the fields of conceptual modeling, database systems, and artificial intelligence. In this paper we present TSOS, a system for reasoning about time that can be integrated as a time expert in environments designed for broader problem-solving domains. The main intended goal of TSOS is to allow a user to infer further information on the temporal data stored in the database through a set of deduction rules handling various aspects of time. For this purpose, TSOS provides the capability of answering queries about the temporal specifications it has in its temporal database. Distinctive time-modeling features of TSOS are the introduction of temporal modalitites , i.e., the possibility of specifying if a piece of information is always true within a time interval, or if it is only sometimes true, and the capability of answering about the possibility and the necessity of the validity of some information at a given time, the association of temporal knowledge both to instances of data and to types of data , and the development of a time calculus for reasoning on temporal data. Another relevant feature of TSOS is the capability to reason about temporal data specified at different time granularities.