The hydrographic record at Station S in the Sargasso Sea shows the development of a subsurface oxygen maximum within the euphotic zone which must be of photosynthetic origin. Consideration of insolation, heat budgets and available 3He/3H data yields an estimate of the order of 5 M m−2 y−1 for the vertically integrated oxygen production rate. Gas exchange calculations reveal a similar oxygen flux leaving the surface, and examination of respiration cycles below the euphotic zone yields consistent results. Such results point to new production of the order of 50 g C m−2 y−1. By using a more realistic mixed layer-thermocline model (Klein and Coste, 1984) it appears that the flux of nutrients into the euphotic zone is sufficient to support such a production. The pulse-like nature of nutrient injection implied by this model raises the possibility of a spatially variable efficiency of recycling which may account for the disparity between the above observations and the level of new production inferred from 14C and 12N incubation techniques.