Background: Outpatient sampling is used to investigate endometrial pathology. Little is known about practice habits and local failure rates at Chris Hani Baragwanath Academic Hospital in Johannesburg, South Africa.Objective: This study assessed the frequency of samples that showed no or limited histological representation of endometrium, and described demographic and pathological features.Methods: All endometrial sample histology reports from the National Health Laboratory Services at the hospital from 01 July 2013 to 31 May 2017 were retrieved by searching the laboratory’s information system. Clinical variables (age, menopausal state, indication for biopsy, endometrial thickness on ultrasound) and pathological findings (macroscopic amount of tissue, histological diagnosis, microscopic presence of endometrial tissue) were extracted and statistically analysed.Results: A total of 1926 samples were included, 91% of which were submitted for abnormal or postmenopausal bleeding. No endometrium was observed in 25% of samples and 13% showed limited endometrium. Benign diagnoses (86%) were most common, with proliferative or secretory changes, endometrial polyps and endometritis accounting for most of these. Associations between the amount of sample received and the presence of endometrial tissue (p ≤ 0.001) and benign versus malignant diagnoses (p ≤ 0.001) were noted. The greater the endometrial thickness, the greater the likelihood of obtaining more sample (bulky vs scant p 0.001) and making a malignant versus benign diagnosis (p = 0.005).Conclusion: These findings are in keeping with literature outside Africa. Histology reports should be explicit when terms such as ‘inadequate’ or ‘insufficient’ are used, in order to facilitate clinical decision-making.