A different way of looking at the meridional warm water (σθ < 26.8) flux in the South and North Atlantic is proposed. The approach involves the blending of observational aspects into analytical modeling, which allows one to circumvent finding a detailed solution to the complete wind–thermohaline problem. The method employs an integration of the momentum equations along a “horseshoe” path in a rectangular basin that is open on the southern side. The initial model considered involves a northward flowing upper layer, a stagnant intermediate layer, and a southward flowing deep layer. By choosing the integration path to begin at one separation point (the Brazil Current separation from South America) and end at another separation point (the Gulf Stream separation), a rather simple expression for the meridional upper-layer transport (T) is obtained. In this scenario the high-latitude cooling affects the warm water northward transport through its influence on the latitude of the western boundary current separation. The authors find that the combined transport (i.e., the transport induced by both wind and high-latitude cooling) is given by T = ∫ (τl/ρ) dl/(f1 − f2), where f1 and f2 are the Coriolis parameters along the northern and southern separation latitudes (i.e., f2 < 0), and τl is the wind stress along the integration path (l). The amount of high-latitude cooling that causes the deep-water formation in the North Atlantic does not enter this relationship explicitly but it does enter the calculations implicitly (through the position of the separation points that adjust to the cooling). Process-oriented numerical experiments, which were conducted using MICOM, are in excellent agreement with the above formula. Surprisingly, application of the formula to the Atlantic gives a transport of less than one Sverdrup (Sv ≡ 106 m3 s−1), an amount that is insignificant compared to the frequently quoted values (10–20 Sv). This questions the common suggestion that surface Atlantic Water flows northward and sinks in high latitude due to wind and high-latitude cooling alone. The difficulty is resolved when a low or midlatitude conversion of Atlantic Intermediate Water to upper thermocline water (via upwelling) is added (à la Goldsbrough) to the model. This implies that the upwelling needed to balance the deep-water formation in the North Atlantic must occur within the limits of the Atlantic Ocean itself rather than in the Pacific and the Indian Oceans. An additional point of interest is that the inclusion of upwelling does not show that all the upwelled water ultimately sinks in the North Atlantic. Rather, it shows that any amount of upwelled intermediate water (at low and midlatitudes) must be equally split between a flow that exits the northern gyre on the north side and a flow that exits the South Atlantic somewhere along the section connecting the mean position of the Brazil Current separation point and the tip of South Africa.