To determine the prevalence of under-nutrition using brief screening methods and to determine the relation between these results and (1) those of a more standard nutritional assessment and (2) discharge outcomes. Prospective study. 65 (21 males) patients older than 65 years. Sub-acute care facility. The Mini Nutritional Assessment, standard nutritional assessment, 'rapid screen' and discharge outcome. The prevalence of under-nutrition was high, ranging from 35.4% to 43.1%, depending on the screening method used. Compared to the standard nutritional assessment the 'rapid screen' consisting of (1) body mass index <22 kg/m(2); and/or (2) reported weight loss of >7.5% over the previous 3 months and the two-tiered Mini Nutritional Assessment process (at risk subjects (46% of total) further evaluated using standard nutritional assessment) had sensitivities of 78.6 and 89.5% and specificities of 97.3 and 87.5% respectively in diagnosing under-nutrition. Under-nourished patients as identified by the standard nutritional assessment (50.0% (under-nourished) versus 21.6% (nourished); P = 0.017), the two-tiered Mini Nutritional Assessment process (50.0% (under-nourished) versus 21.6% (nourished); P = 0.017) and the rapid screen (56.5% (under-nourished) versus 21.4% (nourished); P = 0.004) were more likely to be discharged to an acute hospital or an accommodation with increased support (poor discharge outcomes) than nourished patients. All screening methods identified patients more likely to have a poor discharge outcome. The highly specific but less sensitive 'rapid screen' may be the best method in facilities with limited resources as it can be easily incorporated into nursing/medical admissions and avoids biochemical investigations in all patients. The more sensitive two-tiered Mini Nutritional Assessment is better if resources permit.