To assess the suitability of self-reported oral health status to predict 5- and 10-year tooth loss without involvement of any clinical measures. Within the population Study of Health in Pomerania (SHIP), 2,776 subjects with 5-year follow-up data and 2,016 subjects with 10-year follow-up examination data were studied. Self-rated oral health was assessed at baseline and related to 5- and 10-year tooth loss. Odds ratios and probability of loss were estimated. Five- and 10-year tooth loss was significantly associated with baseline self-rated oral health. The predictive value of oral health ratings for tooth loss was comparable to that of the CDC/AAP case definition or caries and periodontitis diagnostics. In regression analyses including age, sex, smoking, education, income, and further items collected from questionnaires, odds ratios for dichotomous tooth loss associated with rating of oral health as poor were 3.04 (95% CI: 2.43-3.82) and 2.80 (2.11-3.71) after 5 and 10 years, respectively. Prediction with cut-off probability at 0.25 resulted in sensitivity 85.8% and specificity 44.2% for 5 years, and that with cut-off probability at 0.40 resulted in sensitivity 83.5% and specificity 46.5% for 10 years. Self-reported oral health provides reliable predictions of tooth loss comparable to those assessed by clinical diagnostics. © 2018 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd.