How to select patients for upper gastrointestinal endoscopy is a question of paramount clinical importance given the high prevalence of upper gastrointestinal symptoms (in particular of dyspepsia) and the relapsing nature of these complaints. Indeed, about 1 % [1, 2] to 1.8 % [3] of the population undergo gastroscopy each year. The decision to perform a diagnostic procedure such as gastroscopy must take into account the probability and net benefit of identifying disease; in contrast, the decision not to perform it must take into account the risk of missing significant disease [4] for which early, appropriate treatment can improve outcome.In this issue of Endoscopy, two original papers examine the use of upper gastrointestinal endoscopy with regard to patient age. Patient age has traditionally been an important factor for the clinician in deciding whether or not endoscopy should be done. Indeed, the probability of discovering a clinically relevant lesion, including cancer, at gastroscopy increases with age [5 -11]. A gradual increase in diagnostic yield for gastroscopy from 30 % at under 25 years to 61 % at over 65 years [12] has been reported.The paper by Boldys et al. [13] focuses on the relationship between gastric cancer, patient age, and the presence of alarm symptoms. The authors considered 860 patients who underwent a first gastroscopy at a secondary referral center for symptoms compatible with possible gastric cancer. Gastric cancers, most of which were advanced, were found in 83 patients (9.7 %). Of the patients with gastric cancer, 27 % did not have alarm symptoms, and as many as 24 % of them were younger than 45 years. Gastric cancer patients had more symptoms over a shorter period of time than patients without cancer. This paper reports a rate of gastric cancer which is 4 -8 times greater than the prevalence of this disease in Europe as a whole [14]. The main explanation for this provocative and unexpectedly high cancer rate is patient selection, as the prevalence of gastric cancer in Poland seems to be comparable to that in other European countries. Patient selection is significant in this paper for the following reasons. First, the patient sample under consideration represents only 1.3 % of all patients who had undergone gastroscopy at this center over the 10-year period of the study. The results relate to one among eight centers participating in a national gastric cancer program in Poland. Secondly, the centers were allowed to include no more than 100 consecutive patients per year. Thirdly, patient symptoms were already suggestive of possible gastric cancer. Nevertheless, at least in this highly selected population, one out of four patients with gastric cancer was under 45 years old or did not have any alarm symptoms. The authors therefore emphasize the need for endoscopy even in young patients with no alarm symptoms. This is in contrast to current guidelines [15 -17], and certainly cannot be translated into clinical practice as the patient sample in this study is overselected.The paper by Boulton-J...