Assessments of implantation success have mainly been based on incorrect statistical methods: the binomial approach of percentage of implants not failed-to-date over inserted implants. The survival or life-table approach, instead, adjusts individually for duration of exposure (from insertion until explantation or last recall examination). If published failure percentages are adjusted for mean time of follow-up, which is reported in at least some papers, and a lower 95 per cent-confidence bound on the failure rate is calculated, considerably lower probabilities of one, two or more years of survival will be obtained in many cases. An analysis of 2,292 implants yielded a pre-prosthetic 6-months survival probability between 87 and 91 per cent. The post-prosthetic 2-years survival for pillar-shaped (n = 314 with single tooth replacements) and screw implants (n = 150 with edentulous jaws and one implant, randomly chosen, per patient) may be between 79 and 89 per cent and 79 to 96 per cent, respectively (95 per cent-confidence intervals). In pillar-shaped implants, there is preliminary evidence that ceramic implants and male patients carry a larger risk of failure than metal implants and female patients. With screw implants, only high patient age at implantation may be a risk factor.