The relationship between the biogeochemical major elements P/C/N/O in the world ocean has been recognized as being elusive. Here we demonstrate that this elusiveness is caused by the difficulty of separating waters of different origin. These difficulties are reduced if the potential temperature of the water is considered. We employ data obtained from the World Ocean Circulation Experiment Hydrographic Program of oxygen, nitrate, and phosphate. By stratifying these data with respect to potential temperature, clear regression lines emerge between oxygen and nitrate/phosphate. The correlation coefficients are −0.95 and −0.95, respectively. The advantage of using potential temperature (rather than potential or neutral density) comes about because of the strong dependency of oxygen saturation with respect to temperature. In most of the global ocean, the ratio of O2/P/N is found to be close to the Redfield ratio. The exception is colder waters originating in the nutrient-rich surface water of the Southern Ocean and North Pacific. From the data alone, it is difficult to determine whether the non-Redfield ratios are brought about by different stoichiometry or ocean mixing, but our results put strong constraints on mixing and the biogeochemistry in the global ocean.