Caffeine is present in many dietary substances; its appearance from these sources in human milk has not previously been studied in detail. Fifteen lactating women ingested a known amount of a caffeinated beverage (36 to 335 mg). Simultaneous milk and saliva samples were collected at intervals for the subsequent 12 hours and assayed for caffeine content. Eleven of 15 mothers excreted measurable caffeine in milk. Caffeine was detected by 15 minutes in saliva and milk; peak levels in milk (2.09 to 7.17 micrograms/mL) and saliva (1.24 to 9.22 micrograms/mL) were achieved within 1 hour. Elimination half-lives were 1.3 to 13.5 hours (mean 4.0 +/- 3.7 [SD] hours) for saliva and 1.5 to 14.5 hours (mean 6.1 +/- 4.4 [SD] hours) for milk. Assuming each infant would ingest 90 mL of milk every three hours for 24 hours after maternal ingestion of caffeine, it is possible to estimate potential exposure of the nursing infant to caffeine. The amount of caffeine available for infant absorption ranged from 0.01 to 1.64 mg or 0.06% to 1.5% of the maternal dose. Caffeine was not present in the infants' urine collected for five hours after the first nursing period. The maternal ingestion of a single cup of caffeinated beverage does not appear to present significant doses of caffeine to the nursing infant.