The purpose of this study was to apply the principles of generalizability theory to examine the reliability of scoring rubric used in assessing written performance of 11th grade in Oman and to explore the sources of variance that affect reliability of student's scores and detect the degree of consistency and convergence of students' performance. The sample included fifteen 11th grade students from North Batina Governorate in the Sultanate of Oman. They were asked to write on two independent tasks and the tasks were scored by40 raters. G studies showed that the sources of variance that affect more reliability of scores related to: student task interaction and student-rater- method interaction. D studies showed that the G coefficient and index of reliability for holistic scoring rubric did not reach a 0.80 threshold. Also, the results showed that raters' gender and the type of task had no effect on the G coefficient and index of reliability. Key words: sources of variance, G-studies, D- studies