A new standard for X ray dosimeters is based on noninvasive monitoring of X ray tube voltage with detec tion of practical peak voltage (PPV) [1].The standard requires that PPV is measured using discrete probability distribution of voltage within a given range or PPV calculation using a given algorithm. The contrast PPV voltage is determined within a wide range of X ray tube anode voltages.In practice, X ray tube anode voltage is measured using two or more detection channels. Such channels have different spectral sensitivities. A method for PPV calculation from recovery of absolute peak voltage was reported in [2,3]. This method caused a measuring error that cannot be normalized, which is inconsistent with standard requirements. Therefore, the search for new methods of PPV determination is rather urgent.The method of PPV determination suggested in this work is based on the following principles:-bremsstrahlung signal decay is detected using a linear detector;-the signal is minimized for signal spectrum and X ray phantom contrast depending on X ray tube anode voltage;-bremsstrahlung contrast equivalent voltage (PPV) is calculated from a calibration curve.Absorption curves are detected using a linear GaAs detector located along the X ray beam to provide radia tion filtration [4].The X ray spectrum is recovered in this case using spatial localization of detectors as a line. This gives rise to a set of linear algebraic equations S⋅Ewhere S is nor malized quantum detection efficiency (sensitivity) n of the microdetectors within spectral line m (width ΔE); E _ = (E 1 , E 2 , ..., E m ) T is quantum energy vector in the linear microdetectors; I _ = (I 1 , I 2 , ..., I n ) T is signal vector in the n microdetectors. This set of linear algebraic equations is stochastic. It can be solved using a method of directed minimization [5]. This method is based on comparison between two states of the system: μ 1 and μ 2 . The states are determined by vectors P _ apost. = (P 1 , P 2 , ..., P n ) and P _ apr. = (S 11 , S 12 , ..., S nm )·q j , which determine a posteriori probabil ity of quantum absorption P i = N i /N eff of each photode tector and a priori probability S ij = P j ·Q ij of effective absorption of quantum beam q j in each microdetector. The following notation was used: N i is number of absorbed quanta in microdetector i; N eff. is number of absorbed quanta in all microdetectors; P j is probability of quantum absorption at energy E j in the individual detection chan nel; Q ij is probability of absorption of a quantum with energy E j in detector i; q j is probability of generation of a quantum with energy E j in the bremsstrahlung spectrum.Kulback compared macrostates μ 1 and μ 2 and demonstrated that, according to the mean information criterion, μ 1 had advantages over μ 2 [6]:(1) Substitution of probability gives:(2)A procedure for determination of actual peak voltage from recovered spectral distribution of bremsstrahlung is described. The bremsstrahlung spectrum is restored from absorption curves detected using a linea...