This study was carried out at the Agricultural Training and Research Station, King Faisal University, Saudi Arabia. The purpose was to estimate the optimum plot (quadrat) size, shape and number that can be used in large scale in desert rangelands. The weighed index of soil variability (b) was estimated to be 0.69 indicating that medium to large homogenous exists in the experimental site. As the quadrat size (x) increased, the variance among plots (Vs) and comparable variance (v) increased, but the variance per basic unit (V.), and coefficient of variability (c.v.%) tended to decrease with each increase in quadrat size. The relationship between quadrat size and the expected coefficient of variability was described from the equation C.V. = 102.3 X-o.31. The comparable variance method and maximum curvature technique revealed that a quadrat of 24.0 m2 might be considered the optimum quadrat size. The long and narrow quadrat rectangular shape (1m x 24 m) was the most effective in reducing soil variation. The standard error decreased as number of quadrats (replications) or quadrat size increased, but the rate of decrease was more efficient due to the increase in number of quadrats rather than quadrat size. Four quadrats can be used as the optimum number in rangelands of Saudi Arabia.