The unrealistic accumulation of nutrients below the equatorial upwelling regions, termed nutrient trapping by Najjar et al. [1992], has been a major problem of first-generation oceanic carbon cycle models. Suggestions to eliminate this modeling problem include the consideration of dissolved organic nutrients and improvements to the underlying circulation field. This study shows that neither of these suggestions can be complete without a careful investigation of advection numerics. Using a high-resolution coupled biological-physical model of the North and equatorial Atlantic, it is demonstrated for standard central-difference advection (if a too coarse vertical resolution is used) as well as for upstream advection that numerical artifacts alone can lead to a systematic overestimation (by as much as 50%) of simulated subsurface nutrient concentrations in the equatorial upwelling region. A corollary of these results is that without a critical examination of the underlying advection numerics, nutrient trapping alone can no longer present an unchallenged justification for including dissolved organic matter in biogeochemical ocean models. ¿ 2000 American Geophysical Union