This paper develops a dynamic model of inflation in which discretionary monetary policy interacts with distributional conflict between workers and firms. Unlike the canonical Barro-Gordon framework, inflation is socially costly not only because of volatility but also because it redistributes income when nominal wages adjust sluggishly. Policy makers face time-inconsistent incentives to generate inflation in order to stimulate employment, but also internalize the costs of wage adjustment, while workers attempt to defend their real wage subject to bargaining costs. The interaction between policy incentives, wage-setting frictions, and expectation formation renders the optimal inflation rate time-varying and sensitive to institutional features of the labor market. Inflation may be higher or lower than in the absence of distributional conflict, depending on policy priorities over employment versus real wages, the cyclicality of real wages, and the horizon over which wage contracts are reset. When workers possess perfect foresight, stronger real-wage defense dampens inflation and improves welfare by reducing volatility. When prohibitively high information collection costs result in static expectations, however, the same mechanisms reverse the welfare ranking. The framework nests the standard Barro-Gordon outcome as a special case and connects modern policy debates to classical themes concerning wage bargaining, income distribution, policy credibility, and Kalecki's "threat of the sack." By explicitly incorporating distributional considerations into policy optimization, the paper offers a unified approach to understanding inflation persistence and the political economy of macroeconomic stabilization.