Program testing determines whether its behavior matches the specification, and also how it behaves in different exploitation conditions. Proving of program correctness is reduced to finding a proof for assertion that given sequence of formulas represents derivation within a formal theory of special predicted calculus. A well-known variant of this conception is described: correctness based on programming logic rules. It is shown that programming logic rules may be used in automatic resolution procedure. Illustrative examples are given, realized in prolog-like LP-language (with no restrictions to Horn's clauses and without the final failure). Basic information on LP-language are also given. It has been shown how a Pascal-program is being executed in LP-system proffer.