In order to obtain precise and correct dynamic stress-strain behavior for steel plate, the split Hopkinson (Kolsky) bar method or the one bar method has been adopted as a testing method. In these two methods, a dynamic load transducer is the thin steel bar (s). On the input and output bars, typically two or four strain gages are adhered at the same distance from the end of the bars to detect elastic strains of the bars as dynamic load signal. The bars are usually mounted on simple supports, allowing a little axial elongation of the bars. Then, ball bearings or polytetrafluoroethylene parts are frequently installed between the bars and supports, because the friction between them will affect the quality of the dynamic load signal. On the other hand, only for the one bar method, it is reported that a relatively tight support, neighboring the loading end of the output bar, is effective to reduce an extraordinarily-high initial stress peak on dynamic stress-strain curve. In this paper, some trials have been carried out to find the optimum supporting condition for the output bar loading end in the one bar method. An assembly for a steel plate specimen is connected to an impact block. The other end of the assembly is an extension of the output bar. Usually, from the end of gage length of the specimen plate, there is no output bar support within approximately 650mm, for the present apparatus. This situation is designated as “no support”. At a location of 60mm from the end of the gage length, a simple output bar support is introduced additionally. This situation is called as “simple support”, because the output bar is left on the V-shaped top of the support. Additional upper supporting parts can be installed to the simple support condition. After the installation, a square hole is formed on the top of the support. The output bar touches four sides of the hole. This situation is called as “surrounding support”. In addition, specimen assembly types are also included in experimental conditions. Shapes of obtained stress-strain curves in each experimental condition are mutually compared. Also they are compared with that of a reference curve which has the quality of interchangeable with the curves obtained by the split Hopkison bar method. Two experimental conditions are recommended to detect acceptable stress-strain relationships for steel plate.