In this paper the preliminary findings of a multicentre study on the effects of Acetyl-L-Carnitine on mildly impaired elderly are reported. Statistical analysis was carried out on 236 out of 469 subjects sampled in 42 different Italian geriatric or hospital units. Each subject was treated over 150 days, and a battery of tests (investigating cognitive functioning, emotional-affective state and relational behavior) was administered at the beginning on the treatment and the conclusion of each of its four phases. In the first and the last phases there was a 30 days placebo treatment (aimed respectively to wash-out the effects of previous drug and to assess the residual effects of the treatment), while in the second and the third ones (both 45 days long) the subjects took 1500 mg/day of Acetyl-L-carnitine. Repeated multivariate analysis of variance and of covariance (taking as independent variables phases of treatment, age, gender, etiology and severity of mental impairment, as dependent variables the scores either of each test administered or of groups of items and as covariants the level of depression and the sensitivity to placebo effect) showed that drug treatment significantly increased the effectiveness of performance on all the measures of cognitive functioning and of emotional-affective state and on some scores of the relational behavior. Age resulted significantly influential on cognitive functioning and relational behavior, but not on emotional-affective state. Severity of mental impairment significantly influenced also several measures of cognitive functioning and relational behavior, while less consistent results were shown for gender and etiology of mental impairment. The placebo effect, although significant for some cognitive processes, was lower than that of treatment. There findings suggest that Acetyl-L-carnitine treatment is effective against the outcomes of mental impairment in the cognitive (in particular, for memory functioning and constructional thinking) and emotional-affective domains, while its effects on relational behaviour are less consistent, probably because they are partly biased in the subjective evaluation of caregivers and relatives by factors such as age and levels of mental impairment and depression.