The aim of this study was to estimate the relative efficiency (RE) of an On-Farm Experiment conducted in randomized complete block design (RCBD) compared with a completely randomized design (CRD). An on-farm experiment was carried out at Noralhuda, Gezira, Sudan during season 2010. A set of six varieties of sorghum were evaluated in this study and data was collected on flowering period (FP), number of plant (NP), plant height (PH, cm), head length (HL, cm)), 100 seed weight (SW, g), forage weight (FW, kg) and productivity (P, t/ha). Relative efficiency to RCBD, of on-farm experiment was about 91%, 2.04%, 1.22% , 0.91% , 0.88% , 0.99% and 0.89 for FP, NP, PH, HL, 100SW, FW, and P respectively. The coefficients of variation (CV), coefficient of determination (R2) and relative efficiency (RE) were 7.70%, 78% and 1.12% for RCBD respectively. The relative efficiency of on-farm trials show the same efficiency comparing to RCBD, as can be expected. It would have attained the same sensitivity as in RCBD or CRD with about approximately 100% of replication used. The use of completely randomized design was most suited on farms where blocks might be difficult to be formed and impractical to maintain, it was found that experimental design efficiency compared to RCBD is not too low on-farming experimental.