Triaxial shear tests on cohesive soils involve many sources of error. The membrane compliance, membrane resistance, and variations in cross-sectional area during testing are of significance in evaluating test results, especially when the test is conducted under low stress level conditions. The methods of correction to account for these three sources of error are briefly described and an overall correction procedure based on the soil behavior observed during the testing is developed. It is found from correction results that ignorance in making corrections accounting for the three above-mentioned effects will either overestimate or underestimate shear strength of the soil by as much as 30%.