This paper concerns the problem of the control of linear systems by means of feedback from delayed output, where the delay is known and time-varying. The main advantage of the approach is that it can be applied to systems with any delay bound, i.e. not only small delays. The predictor is based on a combination of finite-dimensional elementary predictors whose number can be suitably chosen to compensate any delay. The single-predictor element is an original proposal, and the class of delays to which the schema can be applied includes, but it is not limited to, continuous delay functions.