This study discusses the nature of adjustment costs, which underlie the dynamic theory of input demand. We examine the implications of the conventional assumption that they are quadratic-symmetric and of various alternatives. A recent rapidly growing literature based on microeconomic data shows that the conventional assumption is inferior. We demonstrate the importance of this new knowledge for predicting macroeconomic fluctuations in employment and investment. We indicate its relevance for constructing general equilibrium simulation models, drawing inferences about the likely impacts of labor-market and investment policies, and analyzing firms' dynamic behavior in factor markets.