The labor share may be declining in the data, but it is often assumed constant in neoclassical growth models (NGM). We assess the quantitative importance of this discrepancy by comparing alternative calibration approaches featuring constant and declining labor shares. We find little difference in model performance. Our results derive from strong general equilibrium effects: while a declining labor share mechanically lowers wage growth, the investment response pushes wages back up. Hence, different models deliver nearly identical paths of macro aggregates. Numerous robustness checks (including a CES production function, different time periods, and calculations of the labor share) reinforce the similarity of performance across model specifications. We conclude that the NGM with a constant labor share is still an appropriate choice to study many standard macro aggregates.