SUMMARYBehaviour problems, dependency levels and use of antipsychotic medication were assessed, using an audit approach, in 104 patients in a geriatric psychiatry continuing care unit. High levels of dependency and behaviour disturbance were found. 37.5% of the population were receiving regular antipsychotic medication. The severity of a cluster of four symptoms (diurnal restlessness, agitation, irritability and active aggression) tended to discriminate this group from those on no antipsychotics, though frequency of problems and total number of problems were also higher in the group on regular antipsychotics. Antidepressants were rarely used, though conversely there was only one case of 'missed' depression. Changes following the audit are described and recommendations for improved monitoring of medication use are made.KEY WORDS-Dementia, audit, antipsychotic drugs, dependency, behaviour problems.The use of drugs to control behaviour disturbance associated with severe dementia is controversial. Antipsychotic drugs remain the first-line intervention though there is concern that they can be used inappropriately because of the difficulty in recognizing depression in severely demented patients. Studies tend to be complicated by methodological problems (Helms, 1985), by a sizeable non-response rate, and by the presence of side-effects despite the relatively low doses used. Schneider et al. (1990) found only seven studies (all inpatient) suitable for inclusion in a meta-analysis, reporting that in only 18% of cases could a clear drug effect be presumed. (Overall drug response 59% vs placebo response 41Y0.) They also report no significant differences between antipsychotic drugs used, though individual studies (Rada and Kellner, 1976;Petrie et al., 1982) suggest that high-potency antipsychotics are more effective than low-potency but have more side-effects. There is, however, no evidence to support the hypothesis that sedative compounds are more effective than non-sedating compounds, which suggests that clinical efficacy is a result of antipsychotic effects rather than the effect of 'blanket sedation'.Although large numbers of hospital beds are occupied by those with advanced dementia, there are few studies of the rates of antipsychotic use in this population (Christopher et al., 1978; Michel and Kolakowska 198 1 ;Gilleard et al., 1983; Mann et al., 1984). The first three found rates of 4248%, but the first two include 'prn' medication and all patients over the age of 65, without stating specifically how many had dementia, so rates are not strictly comparable. Mann et al. (1984) report lower rates (28%). Only Michel and Kolakowska (1981) reported numbers of patients on no psychotropics A number of studies examine specific behavioural problems in dementia (eg Burns et al., 1990;Berrios and Brook, 1985; Cummings et al., 1987;Merriam et al., 1988), but only Reisberg et al. (1987). Swearer et al. (1988) and Teri et al. (1988) give figures for total numbers showing behaviour problems in their (outpatient) study populations....