Objectives The present study attempts to 1) identify optimum time limits of test where maximum test reliability can be obtained, 2) examine the variability by the speed and power indexes of test, 3) present how to choose the estimation method of test reliability considering the power and speed indexes and time limits. Methods and Procedures Sampling: Nine groups of subjects were sampled from homogenious high school students. Tools: Reasoning test (Differential Aptitude Test level-Ⅱ-A,B forms standardized by KI-RBS in 1970) Procedures: Nine groups were given test, allowing different time limits to each group. And then they were instructed to continue to work on the same test free from time limits. Findings 1. Time limits of test has to do with speed and power indexes; that is, speed and power indexes of a certain test can be changed by giving different time allowance to subjects. 2. The difficulty level of test is determined by the intellectual level of subjects. Therefore, reliability of test is changeable by the intellectual level of subjects. 3. Reliabilities of tests can be changed by the time limits allowed to the subjects. 4. Time limits must be determined where test scores of subjects ranges widely. 5. Speed factor is also working in power test due to different difficulty level of items according to individual difference, The longer time limits are allowed to subjects, the more speedness of tests will be lessened. 6. High test reliability can be obtained when test is converted from speed to power test according to the speedness analysis fomula presented by G.C. Helmstadter. 7. Split-half reliability estimate method is not valid and sufficient for the estimation of test reliability when test time limit is so short as measure mainly speed factor under the power test condition. 8. Most test except the pure speed test have both speed and power factors. Because every item of the test has different difficulty level. 9. Time limits of tests must be set where individual's test scores maintain constant. 10. The test score of each subject is largely determined by the interaction between the allowed time limits and the difficulty level of tests by individuals. [test socore of a subject=f(allowed time limitsㆍdifficulty level of tests by individuals)] Discussions 1. The reliability is determined by the various factors, eg, No. of items, subject's educational level, No. of cases used for the estimation of reliability, difficulty level, item distribution by discrimation power, time limits, speed and power proportions etc.. Therefore the reliability of test is mentioned, specifically presenting all the data related to reliability. 2. Time limits must be set where the highest reliability coefficint can be obtatined considering all the determinants of reliability. 3. Reliability estimate method should be selected on the base of test difficulty, speed and power proportions. Limitations of the study This study has some limitation in tool and sample size etc.. And also the various coefficients (reliability, speedness and others,) obtained in the study are only restricted to the relevant subjects and tool. Therefore more intensive and extensive study should be followed for the confirmation and generalization of the findings in the study.