The aim of our study was to identify variables of prognostic relevance for disability pensions (DP) in the register data of the German Pension Fund (GPF) and to use the identified variables to construct a risk index. The study was designed as a case-control study of insurants of the GPF Bund using disability pensioners from 2004-2008 as cases and active insurants as controls. Independent variables were selected from the accumulated register data from 2001-2003. Data of 8,500 men and 8,405 women were analyzed. The strongest predictor of future DP were days of sickness benefits. Men with short-term benefits had 6.1 times higher odds of receiving a DP, while men receiving long-term benefits had even 66.3 times higher odds of receiving a DP. For women, the odds were increased 3.8 and 38.4 times, respectively. The risk index score was calculated by transforming the linear combination of parameter estimators and personal characteristics to values ranging from 0-100. ROC analyses and survival analyses confirmed the prognostic relevance of the index score. Independent samples were used to validate our models. Our results show that the GPF has information which could enable an active strategy to enhance the provision of medical rehabilitation.