The accuracy of the recently reported low-resolution NMR method (Goudappel, G. J. W.; et al. J. Colloid Interface Sci. 2001, 239, 535) for the determination of drop-size distribution in oil-in-water emulsions is evaluated by comparing the NMR results with precise data from video-enhanced optical microscopy. A series of 27 soybean-oil-in-water emulsions, differing in their mean drop size, polydispersity, oil volume fraction, and emulsifier, is studied. Soybean oil is selected as a typical component of food emulsions. The experimental error of our optical procedure for drop-size determination is estimated to be around 0.3 microm, which allows us to use the microscopy data as a reference for the mean drop-size and distribution width of the studied emulsions, with known experimental error. The main acquisition parameters in the NMR experiment are varied to find their optimal values and to check how the experimental conditions affect the NMR results. Comparison of the results obtained by the two methods shows that the low-resolution NMR method underestimates the mean drop size, d33, by approximately 20%. For most of the samples, NMR measures relatively precisely the distribution width (+/-0.1 to 0.2 dimensionless units), but for approximately 20% of the samples, larger systematic deviation was registered (underestimate by 0.3-0.4 units). No correlation is found between the emulsion properties and the relative difference between the microscopy and NMR data. Possible reasons for the observed discrepancy between NMR and optical microscopy are discussed, and some advantages and limitations of the low-resolution NMR method are considered.