Life on Earth vitally depends on the availability of water. Human pressure on freshwater resources is increasing, as is human exposure to weather-related extremes (droughts, storms, floods) caused by climate change. Understanding these changes is pivotal for developing mitigation and adaptation strategies. The Global Climate Observing System (GCOS) defines a suite of essential climate variables (ECVs), many related to the water cycle, required to systematically monitor Earth’s climate system. Since long-term observations of these ECVs are derived from different observation techniques, platforms, instruments, and retrieval algorithms, they often lack the accuracy, completeness, and resolution, to consistently characterize water cycle variability at multiple spatial and temporal scales. Here, we review the capability of ground-based and remotely sensed observations of water cycle ECVs to consistently observe the hydrological cycle. We evaluate the relevant land, atmosphere, and ocean water storages and the fluxes between them, including anthropogenic water use. Particularly, we assess how well they close on multiple temporal and spatial scales. On this basis, we discuss gaps in observation systems and formulate guidelines for future water cycle observation strategies. We conclude that, while long-term water cycle monitoring has greatly advanced in the past, many observational gaps still need to be overcome to close the water budget and enable a comprehensive and consistent assessment across scales. Trends in water cycle components can only be observed with great uncertainty, mainly due to insufficient length and homogeneity. An advanced closure of the water cycle requires improved model–data synthesis capabilities, particularly at regional to local scales.