This research was to study the reliability of Modified Essay Questions (MEQ) in Basic Concepts and Principles in Nursing using differrent scoring method, to be used a model of test development for further study. The specific objective of study were as follows 1) To study and compare the reliability of MEQ test by using both Knok and Feletti scoring methods 2) To Study Generalizability Coefficients of MEQ test of each scoring method whether lie within a range of true reliability parameter 3) To study D-study, how many itemss and rater are required to obtain a test reliability of 0.8 at least. The study subject were the 2(nd) nursing students of Nursing College of Ministry of Public Health which consists of 250 students. The research tool used was a model of Basic Concepts and Principles in Nursing, modified by a researcher which contained 4 cases studied, 20 items and 3 raters. Data obtained from the research were calculated for Generalizability and Decision study using GENOVA program (Brenann. 1983). The difference between Generalizability coefficients of Knok and Feletti scoring method were tested by using UX(,I) formula (Woodruff and Feldt.1980). Then, the test reliability of two scoring method were estimated whether lie within a range of true reliability parameter by t formula (Kristof.1972) and Also, the reliabities difference of D study of each scoring methods was tested by ALPHATST program (Lautenschlager.1989). The findings were as follows : 1) Variance components of sum of all items and raters mostly effected on Generalizability Coefficient of two scoring methods. 2) Generalizability Coefficients of Knok scoring method was higher than those Feletti scoring scoring method and significantly different by statistical analysis at p-value 0.01. 3) Generalizability Coefficients of two scoring methods of test lie within a range of true reliability parameters. 4) If a Generalizability Coefficient of Knok scoring method was 0.8. Then, the raters of 4, 3, 2, 2, 2 and 1 were required for the items of 10, 15, 20, 25, 30 and 35 respectively. If a Generalizability Coefficient of Feletti scoring method was 0.8. Then, the raters of 6, 4, 3, 3 and 2 were required for the items of 15, 20, 25, 30 and 35 respectively. And D-study of Feletti scoring method with a Generalizability Coefficient of 0.8 and 1 raters can not be practically applied. 5) It 1-2 raters used for 5 items and 1 rater used for 10 items, it was found that there is no statistically significant difference. It 3-6 raters used for 5 items, 2-6 raters used for 10 items and 1-6 raters used for 15-35 items with each scoring methods, it was found that there is a significant difference between the reliability of two scoring methods at p-value of 0.01.