The research articles published in the "Journal of Research in Science Teaching" in 1965, 1975, 1985, 1995, and in 2005 were surveyed to discover the extent to which they were theory driven. Carey and Smith's theory of the development of science epistemologies was used to frame the study. Specifically their theory posits that science epistemologies develop through three developmental levels. Persons at Level 1 view science as an inductive and descriptive enterprise. Persons at Level 2 view science in terms of hypothesis generation and test. Persons at Level 3 see science as theory driven. That is, theories are generated and their postulates are tested via planned tests with predicted consequences and theories are used to generate specific hypotheses, which in turn are tested in a similar manner. Three computer-generated surveys were conducted on each article identifying articles that contained the words theory, hypothesis, and prediction. The surveys assumed that if the reported research was guided by Level 3 epistemology, then these key terms would be likely to appear somewhere in the article. The term theory was found in only 18.0% of the 1965 articles. Percentages increased in subsequent decades, that is, to 36.5% in 1975, to 44.3% in 1985, to 58.3% in 1995, and then to 86.7% in 2005. Similar, though less dramatic, increases were also found for the terms hypothesis and prediction. Relatively few articles contained two or all three of the terms. Nevertheless, analysis of all 45 research articles appearing in 2005 revealed that most authors were generating and testing hypotheses and/or theories (presumably guided by Levels 2 or 3 epistemology) albeit in a largely implicit and sometimes inconsistent way. Suggestions for improving the quality of research and its reporting are made. (Contains 3 tables and 6 notes.)