The aim of this prospective study was to characterize an implant patient population exhibiting clinical signs of peri-implantitis and to determine subsequently the incidence of progressive attachment loss. The predictive values of diagnostic parameters were evaluated. 25 patients with 54 endosseous implants that had been loaded for 41 +/- 15 months were included in the study. Clinical parameters included the assessment of plaque, bleeding on probing, probing depth, attachment levels, and Periotest values. Probing measurements were performed in duplicate by means of a controlled force electronic probe (Periprobe). Peri-implant crevicular fluid samples were collected and assayed for neutral proteolytic enzyme (NPE) activity (Periocheck). Analysis of duplicate baseline probing data revealed a high degree of reproducibility (mean difference: 0.1 +/- 0.3 mm). A minimum threshold of 1.0 mm (> 3 x S.D.) loss of probing attachment was chosen to classify a site as positive for breakdown. Alternatively, the tolerance method was employed to identify sites with progressive attachment loss. After 6 months, irrespective of the analytical method, 6 percent of all sites (in 19% of the implants) and 28% of the patients had experienced further peri-implant attachment loss. There were significant differences (p < 0.05) in mean plaque (73% vs. 45%) and NPE (36% vs. 12%) scores between patients with progressive peri-implantitis and those with stable peri-implant conditions. Both bleeding on probing and the NPE-test were characterized by high negative predictive values, and thus negative scores can serve as indicators for stable peri-implant conditions. For monitoring peri-implant health during recall visits, attachment level recordings with a controlled force electronic probe in conjunction with enzymatic diagnostic tests of the host response can be recommended.