Extreme precipitation is a major issue for regional climate, hydrology, and safety of society. Our knowledge of extreme precipitation is poor because of difficulties in gauge observations and biases in regional and global datasets, in particular over the snow dominated regions. Here we investigate and report the distribution and magnitude of under-measured of the maximum daily total precipitation (herein daily maximum precipitation) due to biases in manual gauge observations in the high latitudes (over 45°N), using historical data during 1973-2004. We find remarkable patterns in under-measured of the long-term mean daily maximum precipitation and their association to regional climatic regimes. In contrast to relatively small and large-scale under-measured (<5 mm) of daily maximum rainfall, the biases in daily maximum snowfall are very serious, with the regional high values over 15 mm along the Ural Mountains and the coasts of east Asia, Greenland, in particular northern Eurasia coasts. The frequency distribution of observed daily maximum snowfall underestimate significantly the higher risk events over the high latitudes. Furthermore, defining the phase of extreme precipitation should be cautious over the northern regions, in particular the coasts. These results clearly demonstrate the urgent need to review and update precipitation datasets including recent automatic gauge observations and the knowledge of climate regimes and extremes over the broader northern regions.