A cell is the smallest unit of a freely living system. Our understanding of cells is measured by our ability to predict cellular behaviors in response to environmental changes and genetic manipulations. Traditionally, researchers strive to gain insights into cellular behaviors through characterizing the underlying molecular interactions. This ‘bottom-up’ approach however requires a macroscopic number of largely inaccessible parameters in order to be predictive. I will describe a complementary ‘top-down’ approach based on quantitative phenomenology. Extensive quantitative experiments establish that the model bacterium E. coli organizes many of its behavioral responses in very simple manners in accordance to the rate of cell growth. The existence of these simple empirical relations (growth laws) despite myriads of complex molecular interactions is a striking manifestations of a tremendous degree of dimensional reduction occurring in living cells. I will describe how the growth laws can be used to make accurate predictions of cell behaviors without fitting parameters. I will also discuss how the magical dimension reduction can be accomplished by cells through clever strategy of gene regulation.