The aims of this study were to determine whether the nursing costs of functional disability could be assessed retrospectively and, if so, whether this method was sensitive enough to show the influence of a treatment on economic costs. Four studies were available for this purpose: three randomized, double-blind, placebo-controlled, clinical trials measuring, among others, the influence of piracetam on activities of daily living (ADL) of elderly demented patients in Italy, Austria, and Belgium, and one large population-based survey measuring the costs of providing basic nursing, hygienic, and therapeutic needs to 6000 elderly according to ADL scores. The items on the different ADL scales used in each trial were recoded to a common denominator, and an overall dependency score was calculated for each patient. The survey data showed a highly significant linear correlation between ADL status and nursing expenses. Nursing costs of noninstitutionalized patients were calculated by assigning each patient the average nursing cost of all persons from the survey who matched by sex, age, and dependency score. Retrospective economic assessment was possible for ambulatory patients when there were functional assessments and a survey with nursing costs gauged according to those functional assessments. Objective classification criteria for hospital or retirement home patients are needed for the level of reimbursement of nursing costs. Piracetam treatment decreased dependency, compared with placebo. The economic improvement continued over the entire duration of all three clinical trials, lasting up to 6 months in the Italian trial in both ambulatory and residential settings. The decrease in nursing costs remained stable during the same period for patients treated with piracetam, whereas nursing costs of placebo patients steadily increased. After 6 months this difference increased a monthly average of 820 Belgian francs (BF {35 BF = 1 US $}) for ambulatory patients in favor of piracetam, 1650 BF for residential patients, and 2580 BF for hospitalized individuals.