This paper presents a global uncertainty and sensitivity analysis of the Caltech Atmospheric Chemistry Mechanism (CACM). Emphasis is placed on the characterization of uncertainties for product concentrations that constitute secondary organic aerosol (SOA) in CACM. The sensitivity analysis of the chemical mechanism is performed using Monte Carlo techniques combined with Latin hypercube sampling. Uncertainties in rate parameters are propagated through box model simulations with CACM for three different summer cases. Cases studied cover a range of initial concentrations of reactive organic gases and nitrogen oxides that represent episodes of high ozone levels in polluted urban areas. In addition to estimated uncertainties of gas-phase SOA precursor concentrations, similar calculations are performed for O3, HCHO, H2O2, and peroxyacetyl nitrate (PAN). Results indicate that SOA precursor concentrations predicted using nominal CACM rate parameters are similar to estimates from the Monte Carlo simulations. SOA gas-phase precursors in CACM present relative errors that range from 30% at a VOC:NOx ratio of 8:1 to 39% when the ratio changes to 32:1.