Three methods for estimating crown cover of woody vegetation were compared, in three mapped field sites and one theoretical map of randomly distributed individuals, by computer simulated sampling. The effect of plant size on the performance of the methods was investigated by varying the size of individuals in the theoretical map.Using the line-intercept and point-sampling techniques according to standard procedures, the net crown cover of a species is estimated and overlapping areas are not recognized. The variable plot or Bitterlich gauge technique estimates the total crown cover of individuals, and areas of overlap are thus recorded more than once. Consequently, Bitterlich gauge estimates will always be greater whenever individuals intermingle or overtop one another.Line-intercept and point-sampling techniques produced highly variable estimates when plants were small, but were not prone to greater variability when plant distribution departed from randomness. The Bitterlich gauge estimates were no more variable with small than with large plants, but were affected by the degree of plant aggregation. The problem is overcome by selecting a sample point spacing and a halfangle for the gauge that optimizes plant counting.The advantages of the variable plot method over the other two are discussed in terms of speed and operator differences, and it is recommended as a monitoring technique for woody vegetation in arid rangelands.