The Polish Social Insurance Institution (ZUS) is by law obliged to prepare forecasts of the spending and revenues of the Social Insurance Fund (FUS) (i.e. the pension fund, the disability fund, the sickness fund and the work injury fund). A cohort model built to fulfil this requirement was sufficient to forecast aggregate spending and revenues, but, being based on average cohort values, it does not provide information on the benefits’ structure. Therefore a random microsimulation model based on individual data on insured and beneficiaries was built to increase the range of forecasts to assist the decision making authorities in the country.