The rich internal structure and long-range dipole-dipole interactions establish polar molecules as unique instruments for quantumcontrolled applications and fundamental investigations. Their potential fully unfolds at ultracold temperatures, where a plethora of effects is predicted in many-body physics [1,2], quantum information science [3,4], ultracold chemistry [5,6], and physics beyond the standard model [7,8]. These objectives have inspired the development of a wide range of methods to produce cold molecular ensembles [9][10][11][12][13][14]. However, cooling polyatomic molecules to ultracold temperatures has until now seemed intractable. Here we report on the experimental realization of opto-electrical cooling [15], a paradigm-changing cooling and accumulation method for polar molecules. Its key attribute is the removal of a large fraction of a molecule's kinetic energy in each step of the cooling cycle via a Sisyphus effect, allowing cooling with only few dissipative decay processes. We demonstrate its potential by reducing the temperature of about 10 6 trapped CH 3 F molecules by a factor of 13.5, with the phase-space density increased by a factor of 29 or a factor of 70 discounting trap losses. In contrast to other cooling mechanisms, our scheme proceeds in a trap, cools in all three dimensions, and works for a large variety of polar molecules. With no fundamental temperature limit anticipated down to the photon-recoil temperature in the nanokelvin range, our method eliminates the primary hurdle in producing ultracold polyatomic molecules. The low temperatures, large molecule numbers and long trapping times up to 27 s will allow an interaction-dominated regime to be attained, enabling collision studies and investigation of evaporative cooling toward a BEC of polyatomic molecules.The ability to prepare ultracold molecular ensembles has an application potential akin to that of ultracold atoms some decades ago. In fact, the association of KRb dimers [16] as well as the laser cooling of SrF [17] has brought fascinating physics within reach. However, both approaches are restricted to a highly specialized set of purely diatomic molecule species. In order to investigate fundamental physics based on relativistic effects near heavy nuclei or parity violation effects in chiral molecules, or to study molecules of astrophysical, biological, or chemical interest, a more general approach to preparing ultracold molecular ensemble is imperative. This holds in particular for the rich chemical variety of carbon-, nitrogen-, or oxygen-based molecules for which the constituent atoms have not even been laser cooled. Devising a dissipative process to cool such molecules into the ultracold regime has been an exceedingly challenging problem. The standard approach for atoms, laser cooling, is in general impossible for molecules due to the lack of suitable cycling transitions. Creating an artificial cycling transition via cavity cooling [18] has not been demonstrated despite substantial experimental [19] and theoretical [20][21][...