The spatial estimation method of the average of the process over a block whose location and geometry are known is proposed in this paper. A theoretical formulation is presented to estimate conditional lognormal stochastic field when observation is made at some discrete points. The optimum estimator and minimized mean-squared estimation error of a block average value are then compared with the results obtained by Point Kriging when an unknown value at a known location is estimated. Numerical examples were carried out to examine the influence of block size on the optimum estimator and estimated error variance. Except for very small block sizes, the results indicate that the Block Kriging is critical method.