This paper addresses the problems inherent in the design of a guidance filter with heterogeneous sensor outputs, some of which are subject to a known degree of time delay. The proposed method employs an out-of-sequence measurement fusion scheme in the form of fixed-point Kalman smoothing to effectively incorporate the delayed measurements without needing to significantly alter the underlying nonlinear guidance filtering framework. Two numerical case studies concerning satellite proximity operation and anti-ship missile guidance are presented to demonstrate the effectiveness of the proposed method.