Differentiation of sea-salt and non-sea-salt (or excess) components is essential for many studies of marine aerosol and precipitation chemistry. Uncertainties in such calculations arise from (1) uncertainties in the composition of seawater, (2) analytical uncertainties, (3) the amount of dry deposited sea salt in samples, (4) the validity of assuming a purely marine source for the sea-salt reference species, and (5) the validity of assuming no fractionation during or after production of sea-salt aerosols. We assessed these uncertainties and, using the reduced major axis regression technique, evaluated these assumptions by analyzing the composition of precipitation at Amsterdam Island (38¿S, 78¿E) and at Bermuda (32¿N, 65¿W). Precipitation on Amsterdam Island contained significant excess concentrations of Na+ (2%) and Cl- (5%) relative to Mg2+. Indirect evidence indicates that these departures from seawater ratios are probably not due to the influence of terrestrial material. The excess Cl- was sufficient to account for approximately 20% of free acidity, suggesting that scavenging of vapor phase HCl may affect the acid-base chemistry of precipitation at the site. Volume-weighed excess SO42- based on Na+ and Mg2+ differed by 11%, a nonnegligible source of bias in estimating the wet deposition of non-sea-salt sulfur. Precipitation on Bermuda contained significant concentrations of locally derived alkali and alkaline earth metals. The influence of continental and non-sea-salt marine sources on concentrations of excess SO42- was also evident. These observations suggest that assumptions involved in sea-salt corrections are not always satisfied and that it is therefore necessary to evaluate individual data sets, using objective criteria, to select the appropriate reference species.