Silicon isotope tracers were used to examine the relative magnitude of silica dissolution and silica production in the Monterey Bay, California, upwelling system. A diatom bloom dominated by Skeletonema costatum and Chaetoceros spp. was encountered under conditions of moderate upwelling. Profiles of silica production and dissolution rates were obtained at seven stations that sampled both inside and outside the bloom. Integrated silica production rates ranged from 5. ). The mean ratio of integrated silica dissolution to integrated silica production (# D : # P) between the surface and the 0.1% light depth was 0.075, omitting one station with an unusually high # D : # P of 0.61, indicating that, on average, 93% of silica production was supported by new silicic acid. The f-ratio for diatom nitrogen use estimated from silicic acid and nitrate depletion curves and the mean # D : # P ratio was found to be 0.83, indicating that silica was being regenerated at a rate that was only slightly slower than that for particulate organic nitrogen. These data provide direct evidence confirming earlier hypotheses that the silica pump is weak in Monterey Bay. Analysis of the global data set on # D : # P in the surface ocean leads to the hypothesis that low # D : # P (ϳ0.10 or less) are typical of diatom bloom events, with # D : # P rising to values in excess of 0.50 during nonbloom periods. This pattern is shown to be consistent with previous estimates that the annual mean # D : # P ratio in the upper 200 m of the global ocean is 0.5-0.6. A regional analysis reveals that the fraction of silica production supported by new silicic acid varies as a hyperbolic function of the level of gross silica production similar to the variation in the f-ratio for N use with primary productivity. These trends suggest that diatom blooms, especially those occurring in more productive waters, are the main vectors of silica export in the sea, with the majority of the silica produced during nonbloom periods being recycled in the euphotic zone.The relative rates at which silica and organic matter are recycled in the euphotic zone determine whether diatomdominated pelagic ecosystems are driven to silicon or nitrogen limitation. Dugdale et al. (1995) formalized this concept in what they termed the ''silicate pump.'' In their model, grazing and remineralization processes in the euphotic zone tend to recycle organic matter more rapidly than they do particulate silica, causing sinking particles to be enriched in Si over N compared to the relative rates at which particulate organic nitrogen (PON) and biogenic silica (bSiO 2 ) are produced. Thus, the biological pump removes bSiO 2 from the euphotic zone with greater efficiency than it does PON, with the ultimate result that the system is driven toward Si limitation. Here, we use the terminology ''silica pump'' rather than the term ''silicate pump'' because it is biogenic silica, not dissolved silicon, that is exported (i.e., pumped) from the surface ocean to depth. The silica pump seems especially 1 Cor...