For the most important variables in the Portuguese Household Finance and Consumption Survey (ISFF), missing data is imputed using a stochastic multiple imputation algorithm, as agreed in the Household Finance and Consumption Network of the Eurosystem (HFCN). This paper describes the implementation of this methodology in the ISFF. The objective is to get interested readers, namely data users and other producers of survey data, acquainted with one of the most complex and time-consuming stages of the data preparation.