It is assumed that general practitioners can make an important contribution to the diagnosis of dementia. One of the used strategies comprises an evaluation of the Instrumental Activities of Daily Living (IADL). There are contradictory data on the value of this strategy. During one month, 21 Flemish general practitioners evaluated the IADL capacities of all subjects older than 65 years with whom they had contact. Subjects with dementia and/or living in a residential home for the elderly were excluded. Housing and living conditions, medication use and IADL were registered. The general practitioner formulated a clinical evaluation. All subjects with an IADL score > or = 1 and a random sample from the group IADL = 0 underwent a Mini Mental State Examination. Subjects with an IADL score = 4 were referred for neuropsychological and specialist examination. The average age of the 1003 registered subjects was 75.1 years (SD = 6.8). A large majority of them (85%) were totally independent. There was a large discrepancy between the family's and the patient's judgment on the presence of memory problems. There was an inverse correlation between the IADL and MMSE: when the IADL score increased, the MMSE score fell. The diagnostic value of the IADL for the diagnosis of dementia with Camdex-N as a reference standard could not be evaluated because the number of tested subjects was too small. Against the MMSE, sensitivity was 0.81 (SE = 0.03), and specificity was 0.48 (SE = 0.05). The evaluation of the IADL activities had some drawbacks as a detection method for dementia but the use of IADL data may still be clinically valuable in general practice. The correlation between the general practitioner's judgment and that of the specialist was very good. This study showed that the use of the IADL score might change the general practitioner's diagnostic judgment. Furthermore this study confirms the existence of a major threshold for the referral to a specialist of patients with suspected dementia by general practitioners.