Over the last century the Northern Hemisphere has experienced rapid climate warming, but this warming has not been evenly distributed seasonally, as well as diurnally. The implications of such seasonal and diurnal heterogeneous warming on regional and global vegetation photosynthetic activity, however, are still poorly understood. Here, we investigated for different seasons how photosynthetic activity of vegetation correlates with changes in seasonal daytime and night-time temperature across the Northern Hemisphere (>30°N), using Normalized Difference Vegetation Index (NDVI) data from 1982 to 2011 obtained from the Advanced Very High Resolution Radiometer (AVHRR). Our analysis revealed some striking seasonal differences in the response of NDVI to changes in day- vs. night-time temperatures. For instance, while higher daytime temperature (Tmax) is generally associated with higher NDVI values across the boreal zone, the area exhibiting a statistically significant positive correlation between Tmax and NDVI is much larger in spring (41% of area in boreal zone – total area 12.6 × 106 km2) than in summer and autumn (14% and 9%, respectively). In contrast to the predominantly positive response of boreal ecosystems to changes in Tmax, increases in Tmax tended to negatively influence vegetation growth in temperate dry regions, particularly during summer. Changes in night-time temperature (Tmin) correlated negatively with autumnal NDVI in most of the Northern Hemisphere, but had a positive effect on spring and summer NDVI in most temperate regions (e.g., Central North America and Central Asia). Such divergent covariance between the photosynthetic activity of Northern Hemispheric vegetation and day- and night-time temperature changes among different seasons and climate zones suggests a changing dominance of ecophysiological processes across time and space. Understanding the seasonally different responses of vegetation photosynthetic activity to diurnal temperature changes, which have not been captured by current land surface models, is important for improving the performance of next generation regional and global coupled vegetation-climate models.