The paper addresses the use of small-sized specimens of various types, including those with deep (50%) side grooves, for the purpose of fracture toughness prediction. The experimental data for numerous (more than 500) small-sized specimens prepared from materials of various degrees of embrittlement are compared to the test results for full-sized specimens of Ñ(Ò) type. The concepts of Master Curve and Unified Curve are applied for the processing of experimental data. To handle the test results for small-sized deep-grooved specimens a calculation procedure has been elaborated, which adjusts the calculation method specified in the ASTM Standard Å 1921. We provide recommendations of how to use precracked Charpy type deep-grooved (50%) specimens for prediction of a representative temperature dependence of fracture toughness.Introduction. Direct knowledge of the temperature dependence of fracture toughness for a structural steel represents a date source needed for theoretical assessment of reactor pressure vessel integrity during the operation of a nuclear power plant. The applicable standards and codes [1, 2] call for a systematic inspection of the material critical properties within mandatory reference-specimen programs for reactor pressure vessels. In most cases, Charpy V-notch specimens for bending impact tests are used as reference specimens for the determination of brittle fracture resistance. Fracture toughness is usually determined based on the use of precracked Charpy specimens.At present, several methods are available for predicting the temperature dependence of fracture toughness for reactor pressure vessels; noteworthy among them are the Master Curve (MC) method [3,4] and the Unified Curve (UC) method [7,8], which were proposed based on a probabilistic model for the fracture toughness prediction [5,6]. The last-mentioned method allows for a change in trend of the K T Jc ( ) curve with increasing degree of embrittlement of a material. Both methods predict the K T Jc ( ) function for any preset fracture probability and specimen thickness by using test data on standard specimens. The requirements for these tests as well as the procedure of conversion for a preset fracture probability P f = 0 5. and reference specimen thickness B = 25 mm, if the test data were obtained on specimens of arbitrary thickness, are as per ASTM Standard E 1921-02 [3]. This standard implies usage of side-grooved specimens with a groove depth up to 25% of the specimen overall thickness (recommended groove depth is 20%). In practice, tests are performed both on the side-grooved specimens with a groove depth of 20%, as well as on the specimens with no grooves.