The technical objective of this research was to develop a computer method for arranging a number of individual task patterns, representing job incumbents in a given occupational area, into groups or clusters. This advanced computerized technique for clustering work tasks produces homogeneous clusters of task patterns using an input of tasks performed in a sample of jobs. These clusters represent the occupational specialties that exist in a field of work. The important features of this technique are: (1) its capacity for computer analysis of task patterns of large numbers of subjects, (2) its capability for computer assistance in making research decisions at various levels of task analysis, and (3) its flexibility as a tool of pattern recognition and structuring. With only minor modification, the computer programs and concepts described in this report should be of interest to those concerned with other clustering, classifying, and taxonomic techniques. (CH)