The validity and reliability of the Yugoslavian (Beograd) version of the Hungarian adaptation of the Torrance Divergent Capacities Test (HAT-DAT) were tested, with a view toward improving the methodology of scoring the creative abilities test and determining standards for Yugoslavia. The test, based on the work of J. P. Guilford (1977), examines the basic divergent problems of fluency, flexibility, and originality. It contains four subtests--circles, kidney (picture construction), tin can (unusual multiple utility), and monkey (object improvement)--out of the nine in the original Torrance version. The reliability of independent expert assessment of children's paintings and stories was also assessed. Research began in 1978-79 with 104 children; they were studied for the 4 years following enrollment in grade 1. Children were tested with the HAT-DAT in grades 1 and 3. The creative value of artwork and writings from grade 3 was independently assessed by experts from representative fields. The HAT-DAT was not reliable when item analysis, factor analysis, and retesting were used. Correlating test results with results from artwork and stories produced by the children showed that the test was neither valid nor predictive. Independent assessments of children's creative products were not congruent and were no better indicators of divergent abilities than was the test. While the testing method can be improved, results will still not be useful. Study results call the Guilford theoretical approach into question and make the theoretical status of fluency, flexibility, and originality problematic. (SLD)