Digital filter suffer from roundoff noise and adder overflows due to finite word length effects. Scaling is an attempt to internal signal levels such that all signals are as large as possible, yet without the occurrence of overflows. Scaling requirements are implemented by the use of transformer. This paper proposes a procedure for scaling wave digital filters to avoid overflow problems and at the same time maximizing the output signal-to-noise ratio. Results indicate that the scaled networks have an improved signal to noise ratio over th unscaled filters under the condition that there be no overflow occuring.