Abstract The statistical language slang was designed to satisfy the need of unido’s Division for Industrial Studies for an efficient, powerful and easy to use package for descriptive time series analysis. The project began in 1978 shortly after a commercially available dbms (adabas) was chosen to manage unido’s statistical data. Owing to the general nature of adabas, it soon became obvious that the retrieval language provided with it was inadequate for the statistical analyses demanded of the system. At the same time, direct interface of the data base with statistical analysis packages, which normally expect sequential input files, became difficult. Today, slang is used to perform directly most of the statistical analyses required in unido and provides a powerful interface between the data base and statistical packages for more complex tasks.