AIM: To compare in vitro the cleanliness of root canal walls in oval-shaped root canals following automated or manual instrumentation. METHODOLOGY: Forty-five oval-shaped single-rooted maxillary and mandibular premolars with straight canals were divided into three groups of 15. Automated canal preparation was performed using Anatomic Endodontic Technology (AET, group 1) and the ProFile system (group 2). Manual instrumentation (group 3) was performed with K-Flexofiles. Irrigation was performed using alternately 5.25% NaOCl and 17% EDTA, followed by rinsing with saline. The roots were split longitudinally into halves and the canals examined at x200 and x400 in a scanning electron microscope. The presence of debris and smear layer was recorded at distances of 1, 5 and 10 mm from the working length using a three-step scoring scale and a 300 mum square grid. Mean scores for debris and smear layer were calculated and statistically analysed for significance (P < 0.05) between and within groups, using the Kruskal-Wallis nonparametric anova and Dunn's tests. RESULTS: At 1, 5 and 10-mm levels the root canals prepared with AET had significantly less surface debris and smear layer on the canal walls compared with canals prepared with ProFile or manual instrumentation. For all three groups significantly lower mean smear layer scores (P < 0.05) were recorded at 5 and 10-mm levels compared with the 1 mm level. Significantly lower mean debris scores (P < 0.05) were also recorded at 5 and 10-mm levels for the AET group whereas no significant differences were found between the three levels for the ProFile and manual instrumentation groups. CONCLUSIONS: Although better instrumentation scores were obtained in canals prepared with AET, complete cleanliness was not achieved by any of the techniques and instruments investigated.