A double-blind placebo trial has been undertaken on 199 elderly patients admitted to an "acute" geriatric assessment ward. Clinical and biochemical assessment was made on admission (0) and at 2, 4, 8, 16 and 24 weeks (after admission). Ninety-four patients were supplemented with vitamin C (200 mg per day) and 105 had placebo tablets. Biochemical assessment included estimations of plasma and leucocyte (buffy layer) vitamin C, plasma folate, vitamin B12, cortisol and total white cell count. Plasma and leucocyte vitamin C levels remained low for several weeks in a substantial proportion of the non-supplemented patients, whereas low levels were virtually eliminated in the supplemented group. The results from this study suggest that the leucocyte vitamin C levels may give some indication of prognosis in this category of patients (ie. "acute" geriatric admissions) as evidenced by: i) the significantly higher mortality rate during the trial period of patients who started with low initial leucocyte vitamin C levels compared with those starting with higher levels, despite similar mean initial "severity of illness scores" between the two groups. ii) the marked trend, amongst placebo subjects, for those commencing the study with higher leucocyte vitamin C levels to fare better, in terms of progression to "well", than those starting with low levels. Amongst subjects starting with low leucocyte vitamin C levels, there was a trend for "vitamin C" subjects to have fared better by 8 weeks than "placebo" subjects. This again occurred despite similar mean initial "severity scores" between the two groups. Amongst subjects diagnosed with respiratory infections there was some tendency for supplemented patients to fare better than unsupplemented patients. Low leucocyte vitamin C levels, on admission, appear to be predictive of poor subsequent prognosis in elderly hospitalised patients. Results from this trial suggest that supplementation with a moderate dose of vitamin C may improve this prognosis and larger trials with greater numbers appear to be merited to confirm or deny this hypothesis.