Two-field experiments were carried out at the Agricultural Experimental Station of Benha. Kalyobia Govemorate. during two successive seasons (200312004 and 200412005). The aim of this investigation was to determine optimum plot size and number of replications in fertilization experiments for onion yield trials. The experiment included 16 treatments which were the combinations of two bio-fertilizer levels (with bio-fertilizer and without bio-fertilizer) and eight treatments of mineral and organic fertilization (control. N+P+K. V: N+P+K quantities . V. N+P+K quantities. chicken manure. cattle manure. chicken + cattle manure and compost). A split- plot design with three replications were used. Sub plot units were 8 m2. The total onion yield data were recorded for each plot (kgfplot). Data were used to estimate the convenient plot size. and the suitable number of replications using Smith‘s and maximum curvature   methods. Statistical analysis revealed the following results: 1- The index of soil heterogeneity (b) was —0;652 and - 0.534 for the two seasons of experimental. respectively. with an average of —0.593 which reflected intermediate variability in the soil. 2- increasing plot size decreased variance per basic unit and coefficient of variability. However. the reduction was not in proportion with the increase in plot size. 3- The rate of reduction in variance per basic unit and coefficient of variability (CM) decreased as the plot size become larger. 4- The relationship between coefficient of variability (c.v.) and plot size (x) were mathematically expressed b the following equations: C.V. = 23.578 x '°‘ 5 for the first season.   c.v. = 20.635 x “5‘ for the second season. 5- The Optimum plot size. using Smith procedure. was 14. 968 m2 and 9.168 m2 in the first and second seasons. respectively while it in the first and second seasons. respectively using maximum curvature technique wa5213.549m2 and 19.308m2. 6- The results indicated that increasing either number of replications or number of basic units reduced the magnitude of variability between plots. The reduction of difference with increasing number of basic units were less than that obtained by equivalent increase in number of replications.