The p-aminobenzoic acid (PABA) test has been successfully used as an indicator of completeness of 24-h urine collection in field studies of the general population. Our study was designed to investigate its validity for elderly people. Urinary excretion of fractionated oral doses of PABA was measured in 21 young control subjects (19-39 yr old) and 356 elderly (60-89 yr old) men and women. PABA excretion over 24 h was lower in elderly than in control subjects. Subjects aged greater than or equal to 70 yr had a lower recovery of the PABA dose than subjects aged 60-69 yr over the first 24 h, followed by a higher recovery over the next 24-48 h. The cumulative 48-h recovery was similar in all age classes of elderly subjects. However, 48% of the elderly subjects had a cumulative PABA recovery below the conventional cutoff for completeness (85%). These subjects also had consistently lower creatinine output and urinary volume. The lower 24-h urinary PABA recovery over 70 yr of age is interpreted to reflect the delayed renal clearance of the marker substance and indicates that the PABA test is unsuitable for this age group. The low 48-h cumulative recoveries found in all age classes of the elderly are thought to be caused by small unreported losses, which are recurrent in free-living populations.