We present a new, simple, and fast computational technique, termed the incremental slope (IS), that can accurately distinguish between deterministic from stochastic systems even when the variance of noise is as large or greater than the signal, and remains robust for time-varying signals. The IS method is more accurate than the widely utilized Poincare plot analysis especially when the data are severely contaminated by noise. The efficacy of the IS is demonstrated with several simulated deterministic and stochastic signals.