This paper describes DATASIM, a comprehensive software package which generates simulated data for actual or hypothetical research designs. DATASIM is primarily intended for use in statistics and research methods courses, where it is used to generate "individualized" datasets for students to analyze, and later to correct their answers. The generality of its design allows DATASIM to be used to conduct Monte Carlo studies. The methods employed generate datasets whose statistics follow the appropriate sampling distributions. This is true provided that the distribution shape of the variable to be simulated is normal, and that the investigator has not set scale limits on the generation of the data in order to avoid "off range" values. The generation of simulated data by DATASIM for a particular research design is determined by parameters stored in an external file on disk. The application of simulation methods to assess power is examined for several classic studies. The appendices present results of several simulated datasets of the Aronson and Mills (1959) study of the effects of severity of initiation on liking for a group and the Miller and DiCara (1967) study on the instrumental conditioning of heart rate response in rats. (DWH)