This paper investigates inflation forecasting accuracy of several real activity slack measures for the Russian economy. Several Bayesian unobservable-components models using several real activity variables were considered. I show that real-activity slacks gain no improvement in Russian inflation forecasting. This is true for the monthly and for the quarterly data. The estimation was made in the period from the beginning of 2003 to the end of 2018 for monthly data and from the beginning of 1999 to the end of 2018 for the quarterly data. Moreover, their real-times estimates are unreliable in the sense of the magnitude of their revisions.