Living cells have molecular machines for free energy conversion, for example, sliding machines in muscle and other cells, flagellar motors in bacteria, and various ion pumps in cell membranes. They are constructed from protein molecules and work in the nm (nanometer), pN (piconewton) and ms (millisecond) ranges, without inertia. In 1980s, a question was raised of whether the input-output or influx-efflux coupling in these molecular machines is tight or loose, and an idea of loose coupling was proposed. Recently, the long-distance multistep sliding of a single myosin head on an actin filament, coupled with the hydrolysis of one ATP molecule, was observed by Yanagida's group using highly developed techniques of optical microscopy and micromanipulation. This gave direct evidence for the loose coupling between the chemical reaction and the mechanical event in the sliding machine. In this review, I will briefly describe a historical overview of the input-output problem in the molecular machines of living cells.