The aim of the present study was to evaluate the reproducibility of probing depth (PD) and vertical attachment level (PAL-V) measurements at endosseous implants. Twenty partially edentulous patients who had received 44 Frialit-2 and 30 Astra implants between 1994 and 1996 were examined. At all teeth and implants, clinical parameters (GI, PlI, PD, PAL-V) were measured at 6 sites using the flexible plastic version of the universal explorer TPS probe. At the implants and at teeth that were in similar locations compared to the implants, PD and PAL-V measurements were repeated. For each of the 20 probes that had been used for clinical examination of the 20 patients, the probing force was assessed using a precision balance. As a measure of intraexaminer reproducibility, the following standard deviations of single PD and PAL-V measurements were calculated: Frialit-2: 0.71/0.74 mm; Astra: 0.72/0.75 mm; tooth: 0.59/0.57 mm (PD/PAL-V). Stepwise multiple regression analysis revealed implant/tooth position and GI to influence PD and PAL-V measurement error (P<0.001). At anterior teeth, a lower variability was observed than at posterior teeth (P<0.001). PD and PAL-V measurement error were higher at implants than at teeth and influenced by patients. The respective models explained 13% and 17% of the variability of the dependent variable (PD/PAL-V), respectively. Intraexaminer variability of PD and PAL-V measurements at implants tended to be higher than at teeth. Multivariate analysis of variance revealed probe and probe holder to statistically significantly (P<0.001) influence probing force. However, the differences in probing force between the various probe heads were very small. They reached a maximum of 0.054 N between probe 3 and 14 and may be looked upon as clinically irrelevant.