In India, there are different types of higher educational institutions with varying work environment. These variations result in the variation of research productivity. The present study tries to measure the variation in research productivity among teachers belonging to different types of institutions. Primary data on the work environment in terms of contact hours, class strength, regular and occasional responsibilities shouldered by teachers, salary and service benefits along with the background information of the teacher (teaching experience, status of research and subjects taught) were collected from teachers of self financed colleges, aided colleges and Universities in South India. To measure the attitude of the teachers towards research, Attitude Towards Research (ATR) scale was used. It consists of 30 items listed on a 7-point Likert scale. The teacher respondents were classified into three categories - those having high, medium and low levels of attitude towards research. A Composite Research Productivity Index was constructed by considering the number of conferences attended, paper presentations, publication of research papers and the minor and major projects completed by the teacher respondents. Weights were assigned at two levels considering the category of research activities at international, national, state and regional levels and the difficulty level of each research activity. The sum of all these totals constituted Research Productivity Index. This sum was calculated for each individual respondent for further analysis. Linear Multiple Regression was applied to find out the influence of select variables on the research productivity of the respondents. The analysis reveals that a conducive atmosphere in terms of the work environment (teaching work load, teacher-student ratio and other institutional responsibilities) initiate them to engage in research activities as there exist an overall positive attitude towards research which will increase the research productivity of teachers and quality of teaching at large.