Based on practical demands, the rank analysis method of evaluating crop varieties yield in regional trials without replication was proposed. The statistical data of yield in regional trials without replication should match the following supposition: (1) every environment point in regional trials has the equal importance and no difference in weight. (2) The environment points of same trials match unique difference (provided variety) principle; (3) In the environment point of same trial, the environment of trials has only random error and obeys the normal distribution. The quotient of standard deviation (Sv)and theoretical standard(σv)deviation of samples are defined as level of significant difference (LSD) among varieties. This method has proved to be reliable, practical and forthright by the concrete example of regional trials of adzuki bean.
