We study the importance of planning horizons for fiscal multipliers in a New-Keynesian model with bounded rationality. We show that, when agents have shorter planning horizons, government spending multipliers are smaller, whereas labor tax cut multipliers are larger. Furthermore, Ricardian equivalence breaks down, and transfer shocks feature a negative multiplier. Results are driven by the cognitive limitations of finite planning horizons that lead agent’s expectations to deviate from the fully rational benchmark. We find larger investment responses, which are more in line with empirical findings than those of models with longer planning horizons, rule-of-thumb households, or a Blanchard–Yaari structure.