Much interest is currently focused on attempts to estimate aquatic primary production at a variety of spatial and temporal scales using models which require parameterization of phytoplankton photophysiology. The work described here investigates in situ variability in physiological properties which are important for many bio-optical models of photosynthesis: the maximum photosynthetic quantum yield, the pigment-specific absorption coefficient and the product of the effective functional absorption cross section for photosystem II and the rate limiting turnover time for carbon fixation. Measurements were made on samples collected from 2 depths within the euphotic zone at 29 stations in the California Current System (USA) during the winter to early spring of 1992. Within this region, even during a period of relatively low physical forcing, all 3 parameters were found to be as highly variable as previously documented for a wide range of laboratory culture conditions and natural environments. Significant trends with environmental factors such as light, temperature and nutrient conditions were consistent with previous results from controlled laboratory studies of phytoplankton physiology. The consequences of the observed variability for estimation of primary production were investigated through the use of 2 bio-optical models, one expressed in terms of total phytoplankton absorption and the second accounting for only photosynthetically active absorption. Both models were sensitive to the observed variability in physiology, with greater sensitivity to the achieved photosynthetic quantum yield compared to the specific absorption coefficient. Model estimates were significantly less sensitive to observed variability in the absorption coefficient when only the photosynthetically active absorption was included, but little difference in sensitivity to quantum yield variability was found between the 2 models. Despite the documented mesoscale variability in phytoplankton optical and photosynthetic properties, model results suggest that it may be reasonable to use constant but representative parameter values for larger or regional scale estimates of primary production.