Given a known perturbation, distribution function, a problem is formal
ized for constructing a plan which enables the conditions of admissibi
lity and optimality to be fulfilled with the desired probability after
on-line correction of the plan according to the current information o
n the perturbations. This problem of probabilistic planning and the pr
eviously examined problem of guaranteed planning [1] are reduced to a
fern convenient for their comparison. In the general case, properties
of the predominance of probabilistic plans as opposed to guaranteed pl
ans are established. An illustrative example is given to demonstrate t
he agreement between the results of guaranteed planning and those of p
robabilistic planning in the case of unit reliability of the latter pl
anning.