•Patterns of summer forest productivity were examined in the context of winter snow and summer rain.•The GPP season was longer in years with earlier snowmelt.•Total GPP was higher for a longer GPP season in sites with wetter summer climate.•Total GPP was lower for a longer GPP season in sites with drier summer climate.•These patterns were weaker in forests with snow as a higher annual fraction of precipitation. Seasonal snow cover is important in shaping ecosystem carbon uptake across many regions of the world, however forest responses to projected declines in snowpack remain uncertain. We studied the response of forest gross primary productivity (GPP) during the photosynthetically active season to interannual and spatial variability in snow water equivalent (SWE), timing of snowmelt, and length of the active season. We combined carbon flux and weather data from 14 temperate deciduous and evergreen forests in the US and southeast Canada with SWE and precipitation from the Snow Data Assimilation System to test these hypotheses: 1) earlier snowmelt leads to a longer active season; 2) a longer active season is associated with higher total GPP, and 3) GPP during the active season is dependent on peak SWE and timing of snowmelt the previous winter. Regression and correlation analyses did not reveal meaningful environmental predictors of interannual variability in GPP, so linear mixed effects models were used to analyze broader scale spatiotemporal patterns. We found that active season length was negatively correlated with total active season GPP in forests with drier summers on average (based on mean annual summer climatic water deficit), but positively correlated in areas with typically wetter summers. The magnitude of these effects decreased at forests with a higher percentage of annual precipitation falling as snow. Our results showed that the capacity for plants to gain more carbon during a longer active season appears to be dependent on soil water status determined by long-term climate, rather than interannual fluctuations in weather. We found no evidence that the magnitude of total snowfall or peak SWE had a legacy effect on subsequent active season GPP. Finally, we highlight that there was large interannual variability both within and between sites that was not well explained by seasonal climate or phenology.