Biochemical models capable of sustained oscillations and deterministic chaos are investigated. Chaos is characterized by exponential separation of near-by trajectories in the long-term average. However, we observed rather large deviations from purely exponential separation termed "nonuniformity". A quantitative description and consequences of nonuniformity are discussed. Furthermore, the influence of short-correlated noise is treated using next-amplitude maps and Lyapunov exponents. Drastic amplification of fluctuations in non-chaotic systems and relative robustness of chaos were found.