To restore the long term sustainability of pension schemes and improve pension benefits adequacy in recent years many countries have raised the statutory retirement age. According to official reports, however, there is a significant difference between statutory and actual (effective) retirement age. While in some countries the former is considerably higher than the latter, in other it is quite the opposite. Thus a question which underlays this research appears: what determines the effective retirement age? To answer these questions the study objectives are 1) to identify and discuss potential factors affecting retirement decisions; 2) to estimate the impact of the identified factors on the decision to retire. The empirical analysis in the study is based on the crosscountry analysis of 34 OECD countries. To assess the impact of different variables on the effective retirement age we use a multivariate regression model. The model considers variables related to general economic conditions, health, and pension system architecture. Moreover, we also add to a model a dummy variable which informs whether a country is a European one or not. To fit the model and eliminate the highly correlated independent variables we apply the stepwise regression with forward selection. Taking into account the persistent gender differences, we calculate two separate models for men and women. The results of our regression model show that the most significant influence on the retirement age has: perceived health status, life expectancy and the employment rate of people aged 60 to 64. Furthermore, we observe a noticeable impact of "being a European country" variable-simply being the European country leads to 3,76 lower retirement age for men and 2,78 for women. We also find that effective retirement age is positively and statistically significantly correlated with the relative poverty of the elderly, and negatively with old-age dependency ratio and replacement rate. There is also a strong, negative correlation between a dummy "being a European country" variable and effective retirement age. Interestingly, we find no evidence that variables related to pensioners' income (the level of GDP per capita or disposable incomes of people aged over 65 and average wages) affect the effective retirement age. As our model is not capable to fully explain the differences in effective retirement age in analysed countries, we believe that apart from the quantitative parameters regarding economic conditions; financial incentives and pension system architecture; and health and demographic that are considered in the model, also other, rather qualitative factors, influence retirement decisions. We suppose that among these factors are mainly attitudinal and behavioural