PurposeTo analyse the potentiality of STOPP/START criteria for changing inappropriate prescribing (IP) in elderly polypharmacy patients, and their usefulness as perceived by general practitioners (GPs). Subjects and methodsThis was a cross-sectional study with 100 patients aged≥65 years on four medications or more, from 20 GP lists across three health centres. The study variables included: age, sex, comorbidity, medications, IP (STOPP/START criteria), and GP adherence to recommendations, reasons for not adhering and perception of the toolkit's usefulness. Data were collected from electronic medical records and interviews with GPs. ResultsPatients (mean age 77±5.7 years, 64% women) were prescribed a mean of 12.3 drugs/person, 8.7 for chronic conditions. We identified 92 instances of IP in 58 patients (95%CI 48–68%): 55 STOPP criteria in 42 patients (most involving acetylsalicylic acid 20%, NSAIDs 18% or benzodiazepines 16%) and 37 START in 31 patients. For all GPs, ≥1 instance of IP was detected, only two accepting all the recommendations. GPs adhered to 43/92 recommendations (46.7%, 95%CI 36.3–57.1%): 22/55 STOPP (40%, 95%CI 27–53%) and 21/37 START (56.8%, 95%CI 39.5-74.1%). Key reasons for not adhering were not being the prescribing physician (42%) and not seeing benefits (44%). While 95% trusted the recommendations, only 65% thought them feasible. ConclusionsDetecting IP using STOPP/START criteria is no guarantee of improving prescribing to the same extent, since GPs accept<50% of recommendations. While GPs generally appreciate the relevance of the tool and claim to trust it, many believe applying it is not feasible in practice, time being the main barrier cited.