Objective General practitioners (GPs) fail to identify more than 50% of dementia cases using the existing passive case‐finding approach. Using data from the “Ageing in General Practice” study, we sought to establish the additional benefit of screening all patients over the age of 75 for dementia beyond those patients already identified by passive case‐finding. Method Patients were classified as “case‐finding” (n = 425) or “screening” (n = 1006) based on their answers to four subjective memory related questions or their GP's clinical judgement of their dementia status. Cognitive status of each patient was formally assessed by a research nurse using the Cambridge Cognition Examination (CAMCOG‐R). Patients then attended their usual GP for administration of the GP assessment of Cognition (GPCOG) dementia screening instrument, and follow‐up care and/or referral as necessary in light of the outcome. Results The prevalence of dementia was significantly higher in the case‐finding group (13.6%) compared to the screening group (4.6%; p < 0.01). The GPCOG had a positive predictive value (PPV) of 61% in the case‐finding group and 39% in the screening group; negative predictive value was >95% in both groups. GPs and their patients both found the GPCOG to be an acceptable cognitive assessment tool. The dementia cases missed via case‐finding were younger (p = 0.024) and less cognitively impaired (p = 0.020) than those detected. Conclusion There is a very limited benefit of screening for dementia, as most people with dementia could be detected using a case‐finding approach, and considerable potential for social and economic harm because of the low PPV associated with screening.