A one-dimensional model of dimethylsulfide (DMS) oxidation chemistry and simultaneous observations from Flight 24 of the First Aerosol Characterization Experiment are used to test the oxidation of DMS by OH and the gas-phase production of SO2, H2SO4, and methanesulfonic acid (MSA) in the remote marine atmosphere. The model includes a comprehensive chemical mechanism (55 sulfur reactions, 28 sulfur species), vertical mixing, scavenging by background aerosols, and surface losses and emissions. Model parameter uncertainties have been estimated and are used to compute probability distribution functions of observable model outputs using a Monte Carlo method. Seven mechanistic scenarios are considered, which include a baseline case incorporating our best current knowledge and six cases that test novel MSA production reactions involving a newly proposed MSA isomer and the oxidation of methanesulfenic (MSEA) and methanesulfinic (MSIA) acids. The results show that for each of the seven scenarios the modeled DMS and SO2 concentrations agree statistically with the Flight 24 observations. For MSA, however, the observations are a factor of 104 to 105 larger than the baseline mean model predictions and lie 3--4 orders of magnitude outside of the one-sigma model uncertainty range. Statistical agreement between the boundary layer MSA observations and the model is achieved only for the mechanism scenarios that invoke new MSA production pathways, with the best agreement occurring when MSA is produced from the oxidation of MSIA or through a path involving the DMS-OH adduct and MSA isomer. For H2SO4 this study finds that even though the majority of the observations lie within the one-sigma model uncertainty range, the baseline scenario systematically underproduces H2SO4 in the boundary layer. These systematic differences are removed when the production of SO3 is enhanced through a pathway that is independent of SO2. This provides evidence for an efficient H2SO4 production pathway that does not involve SO2 as a precursor. Sensitivity studies are also presented, the results of which suggest observables that are most effective at distinguishing between our seven DMS mechanistic scenarios.