Nectar was collected from the extrafloral nectaries of leaf stipels and inflorescence stalks, and phloem sap from cryopunctured fruits of cowpea plants. Daily sugar losses as nectar were equivalent to only 0.1-2% of the plant's current net photosynthate, and were maximal in the fourth week after anthesis. Sucrose:glucose:fructose weight ratios of nectar varied from 1.5:1:1 to 0.5:1:1, whereas over 95% of phloem-sap sugar was sucrose. [(14)C]Sucrose fed to leaves was translocated as such to nectaries, where it was partly inverted to [(14)C]glucose and [(14)C]fructose prior to or during nectar secretion. Invertase (EC 3.2.1.26) activity was demonstrated for inflorescence-stalk nectar but not stipel nectar. The nectar invertase was largely associated with secretory cells that are extruded into the nectar during nectary functioning, and was active only after osmotic disruption of these cells upon dilution of the nectar. The nectar invertase functioned optimally (phloem-sap sucrose as substrate) at pH 5.5, with a starting sucrose concentration of 15% (w/v). Stipel nectar was much lower in amino compounds relative to sugars (0.08-0.17 mg g(-1) total sugar) than inflorescence nectar (22-30 mg g(-1)) or phloem sap (81-162 mg g(-1)). The two classes of nectar and phloem sap also differed noticeably in their complements of organic acids. Xylem feeding to leaves of a range of (14)C-labelled nitrogenous solutes resulted in these substrates and their metabolic products appearing in fruit-phloem sap and adjacent inflorescence-stalk nectar. (14)C-labelled asparagine, valine and histidine transferred freely into phloem and appeared still largely as such in nectar. (14)C-labelled glycine, serine, arginine and aspartic acid showed limited direct access to phloem and nectar, although labelled metabolic products were transferred and secreted. The ureide allantoin was present in phloem, but absent from both types of nectar. Models of nectary functioning are proposed.