Results from a study on health-related lifestyles of Munich citizens aged between 55 and 65 were compared with an independent data set from Bern. Cluster analysis identified eight health-related behavior patterns in the Munich sample which were also found in the Bern sample. Predictors of health-relevant behavior patterns were analysed via multiple logistic regressions. Most results showed different effects on cluster membership between the two samples. Gender, professional status, social support from friends, and paying attention to regular meals were identified as predictor variables in both samples. Other variables identified as predictors occurred either in only one of the two samples or differed in the direction of their effect.