The early and recent history of meta-analysis is outlined. After providing a definition of meta-analysis and listing its major characteristics, developments in statistics and research are described that influenced the formulation of modern meta-analytic methods. Major meta-analytic methods currently in use are described. Statistical and other research developments contributing to meta-analysis include the introduction of combined tests, combined treatment effects, use of percentages as outcome variables, and use of correlations as outcomes. Meta-analytic approaches reviewed include Glass' methodology, Hedges' modern statistical methods, Hunter and Schmidt's validity generalization, and Rosenthal's methods. Problems affecting meta-analysis include inflated sample sizes, non-independent measures in statistical analyses, the failure to take experimental design into account when estimating effect sizes and sampling errors, and the development of inappropriate statistical methods for testing the influence of study features on study outcomes. Four tables and two graphs are included. (TJH)