OBJECTIVES: To assess the utility of asking for presence of progressive forgetfulness (PF) prior to administering the Abbreviated Mental Test (AMT) when screening for dementia in the community-dwelling elderly Chinese. METHODS: This was a two-phased community-based survey of elderly subjects (>or=50 yrs). In phase one, subjects were asked for PF and administered the AMT. Those having PF or an impaired AMT performance were evaluated clinically for dementia in phase two, which also included a randomly selected sample of 35 subjects with no PF and who passed the AMT. RESULTS: 2,566 subjects completed phase one interview, of which 128 subjects completed phase two. Overall prevalence of PF, failed AMT and dementia were 2.4%, 2.2%, 0.9% respectively. The sensitivity of PF for dementia was 95.7% with specificity of 45.1%. PF was significantly associated with depression in the young-old (50-74 yrs) but not in the old-old (>or=75 yrs) age group, after adjusting for dementia. The probability of subjects (%) in the four possible diagnostic combinations of PF and AMT in the young-old and (old-old) age groups were 0 (0.06) in the no PF/ passed AMT, 0 (0.44) in the no PF/failed AMT, 0.23 (9.2) in the PF/passed AMT and 3.6 (43) in the PF/failed AMT groups. CONCLUSION: In screening for the most common dementias, AMT administration is not required if PF is absent. AMT is also of no added utility for diagnosing dementia in older subjects with PF. Younger subjects with PF should be closely evaluated for depression especially if they passed the AMT, and dementia, if they failed the AMT.