In this paper we introduce MuTACLP, a knowledge representation language which provides facilities for modeling and handling temporal information, together with some basic operators for combining different temporal knowledge bases. The proposed approach stems from two separate lines of research: the general studies on meta-level operators on logic programs introduced by Brogi et al. [7,9] and Temporal Annotated Constraint Logic Programming (TACLP) defined by Frühwirth [15]. In MuTACLP atoms are annotated with temporal information which are managed via a constraint theory, as in TACLP. Mechanisms for structuring programs and combining separate knowledge bases are provided through meta-level operators. The language is given two different and equivalent semantics, a top-down semantics which exploits meta-logic, and a bottom-up semantics based on an immediate consequence operator.