Background: geriatric assessment is a well-established instrument to improve the care of the elderly, but little is known about it in general practice although patients often are known for years. Methods: we used STEP-assessment, an instrument developed by European General Practitioners (GPs), which identifies only problems that can be improved; 37 questions had to be answered by the patient and 4 tests had to be done by the GP. Additionally in the study, GP and patient had to indicate separately which of the problems were seen as relevant and what both accepted to do for improvement. A year later, participating GPs were asked by a not-announced questionnaire to report on improvements and reasons for failure. Results: of the 220 eligible GPs, 45 enrolled a random sample of 894 patients (average age 77 years). In all 7.8 out of 32 possible problems per patient were found. Of those, 1.4 problems were not known to the GP. More than two-thirds of the 'new problems' are perceived as relevant by GP, patient or by both. GPs assessed medical problems and patients assessed social/psychological problems as more relevant. The length and quality of the relationship with the patient was reflected in the number of new problems, with fewer new problems in those well-known. A year later, GPs had offered treatment for 47% of the newly diagnosed problems, with a success-rate of 81%. Conclusion: geriatric screening can detect unidentified problems in general practice. Once detected and dealt with, a high proportion of the undetected problems showed improvement. GPs focussed more on medical, while patients more on psychosocial issues. To increase the outcome of screening, it is necessary to discuss the relevance assessed by the patient.