The transformation of leucine incorporation rates to prokaryotic carbon production rates requires the use of either theoretical or empirically determined conversion factors. Empirical leucine-to-carbon conversion factors (eCFs) vary widely across environments, and little is known about their potential controlling factors. We conducted 10 surface seawater manipulation experiments across the world's oceans, where the growth of the natural prokaryotic assemblages was promoted by filtration (i.e., removal of grazers [F treatment]) or filtration combined with dilution (i.e., also relieving resource competition [FD treatment]). The impact of sunlight exposure was also evaluated in the FD treatments, and we did not find a significant effect on the eCFs. The eCFs varied from 0.09 to 1.47 kg C mol Leu ؊1 and were significantly lower in the FD than in the F samples. Also, changes in bacterial community composition during the incubations, as assessed by automated ribosomal intergenic spacer analysis (ARISA), were more pronounced in the FD than in the F treatments, compared to unmanipulated controls. Thus, we discourage the common procedure of diluting samples (in addition to filtration) for eCF determination. The eCFs in the filtered treatment were negatively correlated with the initial chlorophyll a concentration, picocyanobacterial abundance (mostly Prochlorococcus), and the percentage of heterotrophic prokaryotes with high nucleic acid content (%HNA). The latter two variables explained 80% of the eCF variability in the F treatment, supporting the view that both Prochlorococcus and HNA prokaryotes incorporate leucine in substantial amounts, although this results in relatively low carbon production rates in the oligotrophic ocean. P rokaryotic heterotrophic production (PHP), also known as bacterial production, is a key variable for evaluating the role of prokaryotes in ocean carbon fluxes. However, direct PHP measurements by means of biomass increase with time require long incubations (several days). This procedure is excessively timeconsuming for routine measurements at adequate spatial and temporal scales, and therefore, PHP is typically estimated from related metabolic processes. Determining the incorporation rates of radiolabeled substrates, such as leucine and thymidine, is by far the most widespread approach due to the high sensitivity and the short incubation times required (1, 2). However, the transformation of leucine or thymidine incorporation rates into rates of prokaryotic carbon production relies on the use of conversion factors (CFs). In the case of leucine, a theoretical CF of 3.1 kg C mol Leu Ϫ1 was estimated by Simon and Azam (3) based on the protein content of an average bacterial cell and the typical ratio of carbon-toprotein content, assuming a 2-fold dilution with external leucine (or 1.55 kg C mol Leu Ϫ1 assuming no isotope dilution). Regardless of the systematic application of any of these two theoretical CFs in most published studies, compelling evidence indicates that the relation between leucine ...