Abstract There is presently no satisfactory method of detecting periodontal disease activity at a specified site by means of clinical measurements. This study was designed to examine the possible sources of error with regard to probing measurement reliability. Intra-examiner reproducibility of probing measurements was studied at 766 sites in 10 patients with untreated periodontitis, using a 0.25 N hinged constant force probe (a) with a stent for guidance and landmark, and (b) without stent. The stent made little difference to overall reproducibility of probing depths, though it appeared to reduce variation in different areas. Repeated probing led to an increase in some measurements, perhaps by an effect on tissues. Reproducibility of probing depth was lower in deep pockets, and about 2% of all probing depth scores varied by 3 mm or more at the same site. 4 possible sources of measurement error were noted: visual and tactile observational error, positional error and tissue change. The results are discussed in relation to the clinical detection of periodontal disease activity.