Objective The aim of the present study was to investigate the reproducibility of bacterial enumeration from subsequent subgingival samples collected from patients with peri-implant mucositis. Material and methods Duplicate microbial samples from 222 unique implant sites in 45 adult subjects were collected with paper points and analyzed using the checkerboard DNA-DNA hybridization technique. Whole genomic probes of 74 preselected bacterial species were used. Based on the bacterial scores, Cohen's kappa coefficient was used to calculate the inter-annotator agreement for categorical data. The percentage agreement was considered as "good" when the two samples showed the same score or differed by 1 to the power of 10. Results Moderate to fair kappa values were displayed for all bacterial species in the test panel (range 0.21-0.58). There were no significant differences between Gram-positive and Gram-negative species. The percentage of good agreement between the first and second samples averaged 74.7 % (n=74; range 56-83 %), while the proportion of poor agreement ranged from 1 to 19 % for the various strains. Conclusion While an acceptable clinical agreement was obtained in most cases, diverging bacterial scores may appear in subgingival samples collected at the same time point from patients with peri-implant mucositis. Clinical relevance The broad bulky base of implant crowns may present an obstacle for the collection of reproducible subgingival samples with paper points.