The “green revolution” created dwarf varieties of wheat and rice capable of responding to high fertilizer inputs without lodging. However, this technology relies on fertilizers and has failed to reach resource-poor farmers. Intensive fertilization is not an entirely satisfactory solution in any case, since it causes environmental pollution, and in the case of P, because of limited availability of economically recoverable reserves. With 840 million undernourished people and an increasing global population, a “second green revolution” is required that will increase food production even at low soil fertility, without the need for intensive fertilization. Low soil fertility is a principal constraint to crop production over most of the earth's surface, including nutrient deficiencies as well as toxic levels of Al, Mn, and salt. In the humid tropics and subtropics, where most of the world's poor live, low soil P availability is particularly important due to the cumulative effects of soil weathering and soil reactions that limit P bioavailability. Therefore, the development of P-efficient crop varieties that grow and yield better with low P availability is key to the second green revolution. Substantial genetic variation for P efficiency exists in plants, and breeding programs have been successful in several species. However, these efforts relied on selection for yield in the field, which is slow and is subject to confounding environmental interactions. A better strategy would be to identify and select specific traits that are directly related to P efficiency. Once identified, these traits could be used for phenotypic screening in controlled environments, or tagged with molecular markers for marker-assisted selection. Various root traits could contribute to superior P acquisition. Root exudates including protons, phosphatases, carboxylate anions, and other metal ligands may mobilize P from bound P pools that predominate in many tropical soils. Root morphological traits, particularly root hairs, may be also important for P acquisition. Root anatomical traits that reduce the metabolic cost of soil exploration may also be useful. An important yet poorly understood root trait for P acquisition is root architecture, or the spatial configuration of the root system. Root architecture determines root deployment to distinct soil domains, and therefore determines the accessibility of nutrients to the root system. For example, a shallower root system may be advantageous for P acquisition since P availability is usually higher in the upper layers of the soil and decreases with depth. In common bean, soybean, and maize, plants may produce shallower roots either by developing shallow basal or seminal roots, or by developing strong adventitious root systems. Increased ‘topsoil foraging' through these means may improve P acquisition efficiency by reducing inter-root competition within the same plant and by concentrating root activity in soil domains with the greatest P availability. Other root architectural traits, such as increased proportion of finer root types, may reduce the metabolic cost of soil exploration and therefore increase P efficiency. Genotypes of common bean and soybean with superior root traits have substantially improved growth in low P soils and are now being deployed in breeding programs in Africa, Asia, and Latin America. Although the feasibility of breeding crops with enhanced P acquisition has been demonstrated, the long-term impacts of these genotypes on the productivity of low fertility agroecosystems remains to be determined. Nutrient-extractive genotypes may have altered competition with intercrops, altered nutrient cycling, and altered effects on rotational crops. Root traits that increase P acquisition may have negative consequences for water acquisition. A major benefit from such crops in upland soils should be reduced erosion resulting from greater biomass cover. These issues need to be addressed in future research.