Obesity is a prevalent medical condition in dogs caused by the excess accumulation of fat, with negative effects on quality of life, longevity and the risk of developing associated pathologies. However, it is unclear how frequently first-opinion veterinarians record dogs as overweight (OW) or obese in medical records, and what factors determine when they do. Data sourced through the Small Animal Surveillance Network were used to determine the relative frequency of recording OW status (obesity or OW) in dogs presented to the UK first-opinion practices. Cases were identified using a search of clinical record-free text for relevant keywords. A case–control study was then conducted, comparing dogs where the OW status was recorded with a control group of obese dogs with no diagnosis recorded. Of 49 488 consultations, the OW status was recorded in 671 dogs (relative frequency 1·4 %). Using multiple logistic regression, the OW status of a dog was more likely to be recorded when the consultation was for osteoarthritis (OR 5·42; 95 % CI 2·09, 14·07; P < 0·001) or lameness (OR 2·02; 95 % CI 1·20, 3·42; P = 0·006). Furthermore, the OW status was more commonly recorded in dogs that were members of a practice health scheme (OR 5·35; 95 % CI 1·57, 18·17; P = 0·04) and less commonly recorded in microchipped dogs (OR 0·43; 95 % CI 0·41, 0·91; P = 0·02). These results suggest that OW and obesity are underdiagnosed in the first-opinion practice. However, a presentation for orthopaedic disease appears a key prompt for recording the OW status. Further studies are now warranted to determine the reasons for such marked underdiagnosis.