This study aimed at expanding a new field of educational measurement, by investigating the feasibility of using computer programs for the automatic analysis and evaluation of student writing. Essays written by secondary students in their English classes were rated by multiple independent judges on a number of traits usually considered important: content, organization, style, mechanics, creativity, and overall quality. The essays were key-punched for input to the computer. Computer programs were written to analyze the essays, performing many tests and list lookup procedures, and producing a profile of "proxes" (variables believed to be approximations of important dimensions of the essays). These proxes were then combined through multiple regression to optimize the prediction of the expert judgments. Across various essays, judges, students, and traits, the computer performed about as accurately (in predicting the expert group) as did the typical human judge. Many other dimensions of the problem were examined, including the use of cliches, passive verbs, and syntactic parsing. A plan of attack was outlined for future investigators. (Author)