The carbon isotopic fractionation accompanying formation of biomass by alkenone-producing algae in natural marine environments varies systematically with the concentration of dissolved phosphate. Specifically, if the fractionation is expressed by epsilon p approximately delta e - delta p, where delta e and delta p are the delta 13C values for dissolved CO2 and for algal biomass (determined by isotopic analysis of C37 alkadienones), respectively, and if Ce is the concentration of dissolved CO2, micromole kg-1, then b = 38 + 160*[PO4], where [PO4] is the concentration of dissolved phosphate, microM, and b = (25 - epsilon p)Ce. The correlation found between b and [PO4] is due to effects linking nutrient levels to growth rates and cellular carbon budgets for alkenone-containing algae, most likely by trace-metal limitations on algal growth. The relationship reported here is characteristic of 39 samples (r2 = 0.95) from the Santa Monica Basin (six different times during the annual cycle), the equatorial Pacific (boreal spring and fall cruises as well as during an iron-enrichment experiment), and the Peru upwelling zone. Points representative of samples from the Sargasso Sea ([PO4] < or = 0.1 microM) fall above the b = f[PO4] line. Analysis of correlations expected between mu (growth rate), epsilon p, and Ce shows that, for our entire data set, most variations in epsilon p result from variations in mu rather than Ce. Accordingly, before concentrations of dissolved CO2 can be estimated from isotopic fractionations, some means of accounting for variations in growth rate must be found, perhaps by drawing on relationships between [PO4] and Cd/Ca ratios in shells of planktonic foraminifera.