Abstract: A method of generating the digital representation of a sine wave. A digital signal processor is used to generate the sample digital values, spaced at period T, of a sine wave of frequency F, using the value of the previous sample, e, the corresponding value of cosine, f, and two constants, c=sin2.pi.FT and d=cos2.pi.FT. The next value of sine, S=ed+cf; the value of the corresponding cosine, C=df-ce. The magnitude of c and d are rounded up so that c.sup.2 +d.sup.2 .gtoreq.1. Any calculated values of S or C which exceed unity in magnitude are limited to a magnitude of unity. This limiting operation accomplishes the normalization operation necessary to avoid exponential build-up due to accumulated round-off errors, and does so more efficiently than was previously possible. Sine waves with a 50 dB signal to noise ratio have been generated using this method.